{"title":"NVIDIA GB300 AI Supercomputers","description":"\u003cp\u003eNVIDIA GB300 Grace Blackwell Ultra systems available from SourceIT Singapore, from deskside towers to full liquid-cooled racks.\u003c\/p\u003e\u003cp\u003e\u003cstrong\u003eRack scale:\u003c\/strong\u003e NVIDIA GB300 NVL72 racks from HPE, ASUS, Lenovo and Dell, each with 72 Blackwell Ultra GPUs and 36 Grace CPUs in a single NVLink domain for training and serving the largest AI models.\u003c\/p\u003e\u003cp\u003e\u003cstrong\u003eTower form:\u003c\/strong\u003e NVIDIA DGX Station GB300 workstations from ASUS, Dell, HP, MSI, GIGABYTE and Supermicro, with a GB300 Grace Blackwell Ultra Desktop Superchip and 748 GB of coherent memory for on-premises AI development.\u003c\/p\u003e\u003cp\u003eThese are project systems quoted to your configuration. Email \u003ca href=\"mailto:sales@sourceit.com.sg\"\u003esales@sourceit.com.sg\u003c\/a\u003e or use Request for Quotation on any listing for pricing, availability and lead time.\u003c\/p\u003e","products":[{"product_id":"asus-expertcenter-pro-et900n-g3-gb300-ai-supercomputer","title":"ASUS ExpertCenter Pro ET900N G3 NVIDIA DGX Station GB300 AI Supercomputer","description":"\u003cstyle\u003e\n.pd-acc { border: 1px solid #e2e2e2; border-radius: 8px; margin: 10px 0; overflow: hidden; }\n.pd-acc summary { cursor: pointer; padding: 14px 16px; font-weight: 700; font-size: 1.05em; list-style: none; display: flex; justify-content: space-between; align-items: center; background: #f7f7f7; }\n.pd-acc[open] \u003e summary { border-bottom: 1px solid #e2e2e2; }\n.pd-acc summary::-webkit-details-marker { display: none; }\n.pd-acc summary::after { content: \"+\"; font-size: 1.4em; font-weight: 400; line-height: 1; margin-left: 12px; }\n.pd-acc[open] \u003e summary::after { content: \"\\2212\"; }\n.pd-acc .pd-acc-body { padding: 8px 16px 14px; }\n.pd-spec, .pd-spec tr, .pd-spec td { border: none !important; background: transparent !important; }\n.pd-spec { width: 100%; border-collapse: collapse !important; margin: 6px 0 18px; }\n.pd-spec td { padding: 7px 10px 7px 0 !important; border-bottom: 1px solid #ececec !important; vertical-align: top; font-size: 0.95em; text-align: left; }\n.pd-spec td:first-child { width: 38%; font-weight: 600; color: #333; }\n.pd-spec tr:last-child td { border-bottom: none !important; }\n\u003c\/style\u003e\n\u003cdiv\u003e\n\u003ch2\u003eASUS ExpertCenter Pro ET900N G3 NVIDIA DGX Station GB300 Grace Blackwell Ultra Deskside AI Supercomputer, 748GB Coherent Memory, 20 PFLOPS, Dual 400G ConnectX-8 (ET900N G3) - Local Warranty\u003c\/h2\u003e\n\u003ch3\u003eASUS ExpertCenter Pro ET900N G3 — NVIDIA DGX Station GB300 Deskside AI Supercomputer\u003c\/h3\u003e\n\u003cp\u003eThe ASUS ExpertCenter Pro ET900N G3 is a deskside AI supercomputer built on NVIDIA DGX Station architecture around the NVIDIA GB300 Grace Blackwell Ultra Desktop Superchip — a 72-core NVIDIA Grace CPU (Arm Neoverse V2) paired over NVLink-C2C with an NVIDIA Blackwell Ultra GPU, sharing 748 GB of coherent memory. ASUS rates it at up to 20 petaFLOPS of AI compute, enough to develop, fine-tune and serve large language models of up to 1 trillion parameters locally.\u003c\/p\u003e\n\u003cp\u003eFor a Singapore AI team, research lab, university or enterprise innovation group, the ET900N G3 brings data-centre-class GB300 compute under your own roof and your own data governance — without the lead time, rack space or liquid-cooled rack infrastructure of a full DGX server deployment, and without the recurring cost of cloud GPU instances.\u003c\/p\u003e\n\u003ch3\u003e748 GB Coherent Memory for Trillion-Parameter Models\u003c\/h3\u003e\n\u003cp\u003eThe Grace CPU side carries 496 GB of LPDDR5X at 396 GB\/s and the Blackwell Ultra GPU side carries 252 GB of HBM3e at 7.1 TB\/s — unified into a single 748 GB coherent memory pool over NVLink-C2C rather than two separate memory spaces the software has to shuffle data between. Large models that would not fit in GPU memory alone on a conventional RTX workstation load and run without manual offloading, and Multi-Instance GPU (MIG) can partition the Blackwell Ultra GPU into up to seven isolated instances so several users or workloads share one system.\u003c\/p\u003e\n\u003ch3\u003eUp to 20 PetaFLOPS of AI Performance, Ubuntu with NVIDIA AI Developer Tools\u003c\/h3\u003e\n\u003cp\u003eThe system ships running Ubuntu with the NVIDIA AI Developer Tools stack preinstalled, targeting vLLM, NVIDIA NeMo, NVIDIA NIM and the wider CUDA ecosystem for model development, fine-tuning, RAG pipelines, agentic AI and inference. ASUS has stated that Windows-based AI development support is planned, but Ubuntu is the shipping configuration today.\u003c\/p\u003e\n\u003ch3\u003eEnterprise Storage, Dual 400G ConnectX-8 Networking and Out-of-Band Management\u003c\/h3\u003e\n\u003cp\u003eTwo of the four M.2 2280 NVMe slots run PCIe 5.0 x4 off the Grace CPU and come pre-configured as an OS RAID 1 pair with two 2 TB NVMe SSDs installed; the remaining two slots hang off the ConnectX-8 for training-data storage, for up to 8 TB across all four slots. Three PCIe 5.0 expansion slots (one x16 and two x8) accept an optional NVIDIA RTX PRO Blackwell display GPU or additional NICs. Networking pairs two NVIDIA ConnectX-8 SuperNIC QSFP112 400G ports (transceivers sold separately) with a Marvell 10G Ethernet port and a dedicated Realtek 1G port for the BMC. An ASPEED AST2600 BMC with AMI MegaRAC firmware provides IPMI 2.0 remote management, and an onboard TPM 2.0 handles platform security.\u003c\/p\u003e\n\u003ch3\u003eLiquid-Cooled Deskside Tower with 1600 W Titanium Power Supply\u003c\/h3\u003e\n\u003cp\u003eThe ET900N G3 is a deskside tower measuring 584 × 232 × 565 mm (H × W × D) and weighing 27 kg net. A closed-loop AIO liquid cooling system manages the Grace Blackwell Ultra Superchip's thermal load at approximately 50 dB in Auto Mode at maximum load, and a single 1600 W 80 PLUS Titanium ATX power supply feeds the system. Rated operating temperature is 10 to 35 °C, and the system carries CE, FCC Class A, BSMI, cTÜV SÜD US and KCC certification.\u003c\/p\u003e\n\u003ch3\u003eEnterprise Support — Confirmed at Quotation\u003c\/h3\u003e\n\u003cp\u003eASUS provides onsite service after remote diagnosis for the ET900N G3, with coverage that varies by country; ASUS does not publish a flat multi-year warranty figure for this enterprise-class system. SourceIT confirms the exact ASUS enterprise support tier, term and Singapore service coverage in writing before you order, and supplies every unit with a full GST invoice and local procurement support, including ConnectX-8 transceivers, storage and any rack or power infrastructure your deployment needs.\u003c\/p\u003e\n\u003cdetails class=\"pd-acc\"\u003e\n\u003csummary style=\"cursor: pointer; font-weight: bold;\"\u003eASUS ExpertCenter Pro ET900N G3 Datasheet — Technical Specifications\u003c\/summary\u003e\n\u003cdiv class=\"pd-acc-body\"\u003e\n\u003ch3\u003eGeneral\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\n\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eProduct Name\u003c\/td\u003e\n\u003ctd\u003eASUS ExpertCenter Pro ET900N G3\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eModel \/ Part Number\u003c\/td\u003e\n\u003ctd\u003eET900N G3 (ASUS SKU family: G3-2SSD, G3-4SSD, G3-2SSD-T, G3-4SSD-T, G3-4SSDT-Pro2K)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eCategory\u003c\/td\u003e\n\u003ctd\u003eDeskside AI supercomputer \/ enterprise AI workstation (NVIDIA DGX Station GB300 platform)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eForm Factor\u003c\/td\u003e\n\u003ctd\u003eDeskside tower (not rack-mount)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eManufacturer Support\u003c\/td\u003e\n\u003ctd\u003eOnsite service after remote diagnosis; coverage varies by region — SourceIT confirms current ASUS terms before order\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eLocal Warranty (Singapore)\u003c\/td\u003e\n\u003ctd\u003eProcurement, deployment coordination and GST invoice by SourceIT; ASUS enterprise support terms confirmed at quotation\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\n\u003c\/table\u003e\n\u003ch3\u003eProcessor \u0026amp; GPU\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\n\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eCompute Platform\u003c\/td\u003e\n\u003ctd\u003eNVIDIA DGX Station architecture — NVIDIA GB300 Grace Blackwell Ultra Desktop Superchip\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eCPU\u003c\/td\u003e\n\u003ctd\u003e1 × NVIDIA Grace CPU, 72 Arm Neoverse V2 cores\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eGPU\u003c\/td\u003e\n\u003ctd\u003e1 × NVIDIA Blackwell Ultra GPU (fifth-generation Tensor Cores, second-generation Transformer Engine)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eCPU–GPU Interconnect\u003c\/td\u003e\n\u003ctd\u003eNVIDIA NVLink-C2C, 900 GB\/s (per NVIDIA DGX Station datasheet)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eAI Performance\u003c\/td\u003e\n\u003ctd\u003eUp to 20 PFLOPS FP4 Tensor Core (with sparsity), per NVIDIA — 10 PFLOPS FP8\/FP6, 5 PFLOPS FP16\/BF16\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eModel Support\u003c\/td\u003e\n\u003ctd\u003eAI models up to 1 trillion parameters, per NVIDIA\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eGPU Partitioning\u003c\/td\u003e\n\u003ctd\u003eMulti-Instance GPU (MIG) — up to 7 instances\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\n\u003c\/table\u003e\n\u003ch3\u003eMemory\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\n\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eTotal Coherent Memory\u003c\/td\u003e\n\u003ctd\u003e748 GB unified CPU + GPU memory space\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eGPU Memory\u003c\/td\u003e\n\u003ctd\u003e252 GB HBM3e, 7.1 TB\/s\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eCPU Memory\u003c\/td\u003e\n\u003ctd\u003e496 GB LPDDR5X, 396 GB\/s\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\n\u003c\/table\u003e\n\u003ch3\u003eStorage\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\n\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eOS Storage\u003c\/td\u003e\n\u003ctd\u003e2 × M.2 2280 PCIe 5.0 x4 NVMe (from CPU), pre-configured RAID 1, 2 × 2 TB NVMe SSD installed\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eData \/ Training Storage\u003c\/td\u003e\n\u003ctd\u003e2 × M.2 2280 NVMe slots (from ConnectX-8) for training data, open for expansion\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eMaximum Total Storage\u003c\/td\u003e\n\u003ctd\u003eUp to 8 TB across four M.2 2280 slots, per ASUS\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\n\u003c\/table\u003e\n\u003ch3\u003eExpansion\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\n\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003ePCIe Slots\u003c\/td\u003e\n\u003ctd\u003e3 total — 1 × PCIe 5.0 x16, 2 × PCIe 5.0 x8\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\n\u003c\/table\u003e\n\u003ch3\u003eNetworking\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\n\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eHigh-Speed Networking\u003c\/td\u003e\n\u003ctd\u003e2 × NVIDIA ConnectX-8 SuperNIC QSFP112 ports, 400 Gb\/s each (up to 800 Gb\/s); transceivers not included\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eGeneral Ethernet\u003c\/td\u003e\n\u003ctd\u003e1 × Marvell 10G Ethernet (RJ45)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eManagement LAN\u003c\/td\u003e\n\u003ctd\u003e1 × Realtek 1G Ethernet, dedicated to the BMC\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\n\u003c\/table\u003e\n\u003ch3\u003eI\/O Ports\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\n\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eFront I\/O\u003c\/td\u003e\n\u003ctd\u003e2 × USB 10Gbps Type-C, 2 × USB 10Gbps Type-A, 1 × USB 2.0, headphone and microphone jacks\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eRear I\/O\u003c\/td\u003e\n\u003ctd\u003e4 × USB 10Gbps Type-A, 2 × QSFP112 400G, 1 × 10G Ethernet, 1 × 1G Ethernet (BMC), 1 × mini DisplayPort (BMC), 1 × USB Micro-B (BMC), audio jacks\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\n\u003c\/table\u003e\n\u003ch3\u003eManagement \u0026amp; Security\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\n\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eBMC\u003c\/td\u003e\n\u003ctd\u003eASPEED AST2600 with AMI MegaRAC-based firmware, IPMI 2.0\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eTrusted Platform Module\u003c\/td\u003e\n\u003ctd\u003eOnboard TPM 2.0\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\n\u003c\/table\u003e\n\u003ch3\u003eOperating System \u0026amp; Software\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\n\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eOperating System\u003c\/td\u003e\n\u003ctd\u003eUbuntu with NVIDIA AI Developer Tools preinstalled\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eSoftware Stack\u003c\/td\u003e\n\u003ctd\u003eNVIDIA AI software stack; tested with vLLM; supports NVIDIA NeMo workflows\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\n\u003c\/table\u003e\n\u003ch3\u003ePower, Physical \u0026amp; Environmental\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\n\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003ePower Supply\u003c\/td\u003e\n\u003ctd\u003e1600 W 80 PLUS Titanium ATX; input 115–240 V AC, 15–8 A, 50–60 Hz\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eCooling\u003c\/td\u003e\n\u003ctd\u003eClosed-loop AIO liquid cooling; approx. 50 dB in Auto Mode at maximum load (without discrete VGA)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eDimensions (H × W × D)\u003c\/td\u003e\n\u003ctd\u003e584 × 232 × 565 mm (23.0 × 9.1 × 22.3 in)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eWeight\u003c\/td\u003e\n\u003ctd\u003e27 kg net; 32 kg gross\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eOperating Temperature\u003c\/td\u003e\n\u003ctd\u003e10 – 35 °C\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eCertifications\u003c\/td\u003e\n\u003ctd\u003eCE, FCC Class A, BSMI, cTÜV SÜD US, KCC\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\n\u003c\/table\u003e\n\u003c\/div\u003e\n\u003c\/details\u003e\n\u003cdetails class=\"pd-acc\"\u003e\n\u003csummary style=\"cursor: pointer; font-weight: bold;\"\u003eCompatibility \u0026amp; Software Support\u003c\/summary\u003e\n\u003cdiv class=\"pd-acc-body\"\u003e\n\u003ch3\u003eNVIDIA AI Software Stack\u003c\/h3\u003e\n\u003cp\u003eShips with Ubuntu Linux and the NVIDIA AI Developer Tools preinstalled — the same NVIDIA software foundation used across the DGX family. CUDA, cuDNN, TensorRT and TensorRT-LLM, NVIDIA NeMo, NVIDIA NIM microservices, Triton Inference Server, RAPIDS and NGC containers run natively on the Grace Blackwell Ultra platform, alongside PyTorch, TensorFlow, JAX, vLLM, Hugging Face Transformers, DeepSpeed and ONNX Runtime. Docker and Kubernetes deployment are supported for containerised MLOps pipelines. NVIDIA AI Enterprise is available as a separate subscription for supported production deployments.\u003c\/p\u003e\n\u003ch3\u003eArm64 Architecture — Plan Your Software Accordingly\u003c\/h3\u003e\n\u003cp\u003eThe Grace CPU is an Arm Neoverse V2 processor, so the operating system, drivers and container images are arm64 (aarch64) builds. Mainstream AI frameworks and NVIDIA NGC containers are published for arm64, but proprietary x86-only applications will not run natively — check any commercial software you rely on before ordering. NVIDIA has announced a separate DGX Station for Windows SKU (Windows plus Windows Subsystem for Linux) for a later release; the units SourceIT supplies today ship with Ubuntu.\u003c\/p\u003e\n\u003ch3\u003eClustering and Networking Compatibility\u003c\/h3\u003e\n\u003cp\u003eThe two NVIDIA ConnectX-8 SuperNIC QSFP112 ports run at 400 Gb\/s each (up to 800 Gb\/s aggregate) with RDMA over Converged Ethernet, GPUDirect RDMA and GPUDirect Storage support, so two or more systems can be linked for distributed training, or connected to an NVMe-over-fabric storage array. QSFP112 transceivers or direct-attach cables are ordered separately to match your switch fabric — SourceIT will quote compatible optics alongside the system. The 10GbE RJ45 port covers everyday site networking, and the dedicated management LAN keeps out-of-band BMC traffic isolated from production data.\u003c\/p\u003e\n\u003ch3\u003ePositioning Within the ASUS Range\u003c\/h3\u003e\n\u003cp\u003eThe ET900N G3 sits above ASUS's conventional ExpertCenter desktops, ProArt workstations and NUC \/ PN-series Mini PCs — those are Windows business PCs with integrated NPUs for on-device Copilot+ features. The ET900N G3 is a different category entirely: a Linux-based, GPU-accelerated deskside supercomputer for model training, fine-tuning and inference. If you are unsure which category fits your project, email \u003ca href=\"mailto:sales@sourceit.com.sg\"\u003esales@sourceit.com.sg\u003c\/a\u003e describing the workload and we will help you scope the right system.\u003c\/p\u003e\n\u003c\/div\u003e\n\u003c\/details\u003e\n\u003cdetails class=\"pd-acc\"\u003e\n\u003csummary style=\"cursor: pointer; font-weight: bold;\"\u003ePackage Contents \u0026amp; Ordering Information\u003c\/summary\u003e\n\u003cdiv class=\"pd-acc-body\"\u003e\n\u003ch3\u003eWhat Ships With the System\u003c\/h3\u003e\n\u003cul\u003e\n\u003cli\u003eASUS ExpertCenter Pro ET900N G3 system unit with NVIDIA GB300 Grace Blackwell Ultra Desktop Superchip\u003c\/li\u003e\n\u003cli\u003e2 × 2 TB NVMe M.2 SSD pre-installed and configured as OS RAID 1 (base G3-2SSD configuration; 4-SSD and transceiver-bundled variants available to order)\u003c\/li\u003e\n\u003cli\u003ePower cord for the 1600 W PSU\u003c\/li\u003e\n\u003cli\u003eASUS enterprise system documentation\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003cp\u003eQSFP112 transceivers (unless a -T variant is ordered) and additional NVMe SSDs for the training-data slots are quoted separately to match your networking and storage requirements.\u003c\/p\u003e\n\u003ch3\u003eNVIDIA DGX Station GB300 Systems at SourceIT Singapore\u003c\/h3\u003e\n\u003cp\u003eThe ASUS ExpertCenter Pro ET900N G3 is scoped and quoted individually by SourceIT — we confirm SSD configuration, transceivers, power circuit requirements and the applicable ASUS enterprise support terms before finalising a quotation.\u003c\/p\u003e\n\u003cul\u003e\n\u003cli\u003e\n\u003ca href=\"\/products\/msi-xpertstation-ws300-nvidia-dgx-station-ai-supercomputer\"\u003eMSI XpertStation WS300\u003c\/a\u003e — NVIDIA DGX Station GB300 platform, same 748 GB \/ 20 PFLOPS core compute\u003c\/li\u003e\n\u003cli\u003e\n\u003ca href=\"\/products\/hp-zgx-fury-g1n-gb300-ai-workstation-with-nvidia-blackwell-ultra\"\u003eHP ZGX Fury AI Station G1n\u003c\/a\u003e — NVIDIA DGX Station GB300 platform, same 748 GB \/ 20 PFLOPS core compute\u003c\/li\u003e\n\u003cli\u003e\n\u003ca href=\"\/products\/gigabyte-w775-v10-l01-deskside-ai-supercomputer-with-nvidia-gb300-blackwell\"\u003eGIGABYTE W775-V10-L01\u003c\/a\u003e — NVIDIA DGX Station GB300 platform, same 748 GB \/ 20 PFLOPS core compute\u003c\/li\u003e\n\u003cli\u003e\n\u003ca href=\"\/products\/nvidia-dgx-spark-ai-supercomputer-4tb-940-54242-0007-000\"\u003eNVIDIA DGX Spark\u003c\/a\u003e — desktop-scale GB10 system (128 GB unified memory, 1 PFLOP FP4) for individual developers\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003cp\u003eBrowse the full \u003ca href=\"\/collections\/nvidia-dgx-spark\"\u003eNVIDIA DGX Spark \u0026amp; DGX Station collection\u003c\/a\u003e. For volume, project, research-grant or education pricing, GST invoicing and current lead times, email \u003ca href=\"mailto:sales@sourceit.com.sg\"\u003esales@sourceit.com.sg\u003c\/a\u003e.\u003c\/p\u003e\n\u003c\/div\u003e\n\u003c\/details\u003e\n\u003cdetails class=\"pd-acc\"\u003e\n\u003csummary style=\"cursor: pointer; font-weight: bold;\"\u003eDatasheet \u0026amp; Downloads\u003c\/summary\u003e\n\u003cdiv class=\"pd-acc-body\"\u003e\n\u003cp\u003eOfficial ASUS documentation for the ExpertCenter Pro ET900N G3:\u003c\/p\u003e\n\u003cul\u003e\n\u003cli\u003e\u003ca href=\"https:\/\/dlcdnwebimgs.asus.com\/files\/media\/235482bc-bbbe-4347-b7e9-0969da7f8268\/asus-expertcenter-pro-et900n-g3-datasheet.pdf\" rel=\"noopener\" target=\"_blank\"\u003eASUS ExpertCenter Pro ET900N G3 datasheet (PDF)\u003c\/a\u003e\u003c\/li\u003e\n\u003cli\u003e\u003ca href=\"https:\/\/www.asus.com\/sg\/displays-desktops\/workstations\/performance\/expertcenter-pro-et900n-g3\/\" rel=\"noopener\" target=\"_blank\"\u003eASUS ExpertCenter Pro ET900N G3 — product page (Singapore)\u003c\/a\u003e\u003c\/li\u003e\n\u003cli\u003e\u003ca href=\"https:\/\/www.asus.com\/displays-desktops\/workstations\/performance\/expertcenter-pro-et900n-g3\/techspec\/\" rel=\"noopener\" target=\"_blank\"\u003eASUS ExpertCenter Pro ET900N G3 — full technical specifications\u003c\/a\u003e\u003c\/li\u003e\n\u003cli\u003e\u003ca href=\"https:\/\/www.nvidia.com\/en-sg\/products\/workstations\/dgx-station\/\" rel=\"noopener\" target=\"_blank\"\u003eNVIDIA DGX Station (GB300) — official platform page (Singapore)\u003c\/a\u003e\u003c\/li\u003e\n\u003cli\u003e\u003ca href=\"https:\/\/www.nvidia.com\/content\/dam\/en-zz\/Solutions\/products\/workstations\/dgx-station\/workstation-datasheet-dgx-station-nvidia-us-5543600-web.pdf\" rel=\"noopener\" target=\"_blank\"\u003eNVIDIA DGX Station datasheet (PDF)\u003c\/a\u003e\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003cp\u003eFor deployment guides, current ASUS enterprise support terms or Singapore availability, email \u003ca href=\"mailto:sales@sourceit.com.sg\"\u003esales@sourceit.com.sg\u003c\/a\u003e quoting ET900N G3.\u003c\/p\u003e\n\u003c\/div\u003e\n\u003c\/details\u003e\n\u003cdetails class=\"pd-acc\"\u003e\n\u003csummary style=\"cursor: pointer; font-weight: bold;\"\u003eFrequently Asked Questions — ASUS ExpertCenter Pro ET900N G3\u003c\/summary\u003e\n\u003cdiv class=\"pd-acc-body\"\u003e\n\u003ch3\u003eWhat is the ASUS ExpertCenter Pro ET900N G3?\u003c\/h3\u003e\n\u003cp\u003eA deskside AI supercomputer built on NVIDIA DGX Station architecture around the NVIDIA GB300 Grace Blackwell Ultra Desktop Superchip, with 748 GB of coherent CPU–GPU memory and up to 20 PFLOPS of AI performance for developing, fine-tuning and serving large language models of up to 1 trillion parameters locally. It is not a general office PC or a conventional engineering workstation.\u003c\/p\u003e\n\u003ch3\u003eIs this the same as the NVIDIA DGX Station GB300?\u003c\/h3\u003e\n\u003cp\u003eYes — the ET900N G3 is ASUS's build of the NVIDIA DGX Station GB300 reference platform. The GB300 Superchip, 748 GB coherent memory, 20 PFLOPS FP4 compute, ConnectX-8 networking and NVIDIA AI software stack are the same across every OEM DGX Station; the OEMs differ in chassis, cooling, storage configuration, management firmware and support terms.\u003c\/p\u003e\n\u003ch3\u003eHow does it compare to the NVIDIA DGX Spark?\u003c\/h3\u003e\n\u003cp\u003eDifferent class entirely. The DGX Spark is a GB10 desktop system with 128 GB unified memory and 1 PFLOP FP4 for models up to about 200 billion parameters; the ET900N G3 delivers 748 GB and up to 20 PFLOPS for models up to 1 trillion parameters. Choose the Spark for an individual developer, the ET900N G3 for a team or production-scale AI.\u003c\/p\u003e\n\u003ch3\u003eWhat operating system does it run?\u003c\/h3\u003e\n\u003cp\u003eUbuntu with the NVIDIA AI Developer Tools stack preinstalled. ASUS has indicated that Windows-based AI development support is planned, but Ubuntu is the shipping configuration today.\u003c\/p\u003e\n\u003ch3\u003eCan several people share one system?\u003c\/h3\u003e\n\u003cp\u003eYes. Multi-Instance GPU (MIG) partitions the Blackwell Ultra GPU into up to seven isolated instances, so multiple users or workloads run concurrently with guaranteed resources.\u003c\/p\u003e\n\u003ch3\u003eDoes it need a server room or a liquid-cooled rack?\u003c\/h3\u003e\n\u003cp\u003eNo rack is required — cooling is a closed-loop AIO liquid loop built into the deskside tower. Plan an air-conditioned space within the 10–35 °C operating range and a dedicated power circuit for the 1600 W Titanium PSU under sustained AI load.\u003c\/p\u003e\n\u003ch3\u003eWhat networking does it need, and are transceivers included?\u003c\/h3\u003e\n\u003cp\u003eTwo NVIDIA ConnectX-8 QSFP112 400G ports handle clustering and high-speed storage, with 10G and 1G Ethernet for general and management traffic. Transceivers are not included on the base SKU (ASUS offers -T variants that add two) and must match your switch fabric — SourceIT quotes compatible optics alongside the unit.\u003c\/p\u003e\n\u003ch3\u003eWhat warranty or support does ASUS provide in Singapore?\u003c\/h3\u003e\n\u003cp\u003eASUS states onsite service after remote diagnosis, with coverage varying by country, and publishes no flat multi-year figure for this system. SourceIT confirms the current ASUS enterprise support term and Singapore coverage in writing as part of your quotation.\u003c\/p\u003e\n\u003ch3\u003eHow do I buy the ASUS ET900N G3 in Singapore?\u003c\/h3\u003e\n\u003cp\u003eSourceIT is an authorised IT reseller in Singapore and supplies the ET900N G3 with a full GST invoice, local procurement support and deployment scoping — email sales@sourceit.com.sg with your workload and site details for a quotation and current lead time.\u003c\/p\u003e\n\u003c\/div\u003e\n\u003c\/details\u003e\n\u003ch3\u003eProduct Videos\u003c\/h3\u003e\n\u003cp\u003e\u003ciframe width=\"560\" height=\"315\" src=\"https:\/\/www.youtube.com\/embed\/z70xOVy3V40\" title=\"ASUS ExpertCenter Pro ET900N G3 – Deskside AI Supercomputer built on NVIDIA DGX Station\"\u003e\u003c\/iframe\u003e\u003c\/p\u003e\n\u003cp\u003e\u003ciframe width=\"560\" height=\"315\" src=\"https:\/\/www.youtube.com\/embed\/WNmTKjAAAgs\" title=\"ASUS ExpertCenter Pro ET900N G3 | The Deskside Data Center\"\u003e\u003c\/iframe\u003e\u003c\/p\u003e\n\u003c\/div\u003e","brand":"Asus","offers":[{"title":"Default Title","offer_id":51421362847908,"sku":"ET900N G3","price":170000.0,"currency_code":"SGD","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0605\/0136\/0804\/files\/asus-expertcenter-pro-et900n-g3-agent-computer-v2.png?v=1791980211"},{"product_id":"msi-xpertstation-ws300-nvidia-dgx-station-ai-supercomputer","title":"MSI XpertStation WS300 NVIDIA DGX Station GB300 AI Workstation","description":"\u003cdiv\u003e\n\u003cstyle\u003e\n.pd-acc { border: 1px solid #e2e2e2; border-radius: 8px; margin: 10px 0; overflow: hidden; }\n.pd-acc summary { cursor: pointer; padding: 14px 16px; font-weight: 700; font-size: 1.05em; list-style: none; display: flex; justify-content: space-between; align-items: center; background: #f7f7f7; }\n.pd-acc[open] \u003e summary { border-bottom: 1px solid #e2e2e2; }\n.pd-acc summary::-webkit-details-marker { display: none; }\n.pd-acc summary::after { content: \"+\"; font-size: 1.4em; font-weight: 400; line-height: 1; margin-left: 12px; }\n.pd-acc[open] \u003e summary::after { content: \"\\2212\"; }\n.pd-acc .pd-acc-body { padding: 8px 16px 14px; }\n.pd-spec, .pd-spec tr, .pd-spec td { border: none !important; background: transparent !important; }\n.pd-spec { width: 100%; border-collapse: collapse !important; margin: 6px 0 18px; }\n.pd-spec td { padding: 7px 10px 7px 0 !important; border-bottom: 1px solid #ececec !important; vertical-align: top; font-size: 0.95em; text-align: left; }\n.pd-spec td:first-child { width: 38%; font-weight: 600; color: #333; }\n.pd-spec tr:last-child td { border-bottom: none !important; }\n\u003c\/style\u003e\n\u003ch2\u003eMSI XpertStation WS300 NVIDIA DGX Station GB300 Grace Blackwell Ultra AI Workstation, 748GB Coherent Memory, 20 PFLOPS, 2TB RAID 1, Dual 400G ConnectX-8, Liquid Cooled (WS300T60L) - Local Warranty\u003c\/h2\u003e\n\u003ch3\u003eMSI XpertStation WS300 — NVIDIA DGX Station GB300 AI Workstation for the Deskside\u003c\/h3\u003e\n\u003cp\u003eThe MSI XpertStation WS300 (model WS300T60L) is MSI's enterprise AI workstation built on NVIDIA DGX Station architecture around the NVIDIA GB300 Grace Blackwell Ultra Desktop Superchip. A 72-core NVIDIA Grace CPU and an NVIDIA Blackwell Ultra GPU share 748 GB of coherent memory over NVLink-C2C, delivering up to 20 petaFLOPS of FP4 AI performance and support for models of up to 1 trillion parameters — data-centre-class AI in a liquid-cooled tower that sits beside a desk.\u003c\/p\u003e\n\u003cp\u003eFor Singapore enterprises, research institutes and AI startups, the WS300 is the on-premise alternative to renting cloud GPU capacity: train, fine-tune and serve frontier-scale LLMs, agentic AI and simulation workloads on your own hardware, under your own data sovereignty, with an NVIDIA-certified and Ubuntu-certified platform.\u003c\/p\u003e\n\u003ch3\u003e748 GB Coherent Memory: One Memory Space for Giant Models\u003c\/h3\u003e\n\u003cp\u003e252 GB of HBM3e GPU memory at 7.1 TB\/s pairs with 496 GB of LPDDR5X CPU memory at 396 GB\/s over the 900 GB\/s NVLink-C2C link, so models and datasets live in a single 748 GB coherent memory space instead of being paged between CPU and GPU. That accelerates LLM fine-tuning, retrieval-augmented generation, multimodal inference and scientific simulation, and Multi-Instance GPU (MIG) lets up to seven users or workloads share the Blackwell Ultra GPU with guaranteed resources.\u003c\/p\u003e\n\u003ch3\u003e2 TB RAID 1 Boot Storage, PCIe 6.0 Expansion and 800G-Class Networking\u003c\/h3\u003e\n\u003cp\u003eTwo PCIe 5.0 x4 M.2 2280 slots off the Grace CPU ship populated as a 2 TB RAID 1 boot pair, while two further PCIe 6.0 x4 M.2 2280 slots off the ConnectX-8 remain open for high-speed dataset and checkpoint storage. Dual QSFP112 400G Ethernet ports on the NVIDIA ConnectX-8 SuperNIC (up to 800 Gb\/s aggregate) connect the WS300 into private AI clusters and distributed training fabrics, with a Marvell AQC113 10GBase-T port for site networking and a dedicated 1000Base-T management port for the BMC.\u003c\/p\u003e\n\u003ch3\u003eLiquid-Cooled, Enterprise-Managed, Built for 24 × 7 AI Workloads\u003c\/h3\u003e\n\u003cp\u003eA liquid cooling module rated for up to 1400 W of CPU + GPU load, backed by a 120 mm system fan, keeps the Grace Blackwell Ultra Superchip at full performance under sustained training runs. An ASPEED AST2600 BMC with AMI MegaRAC firmware supports IPMI 2.0 and the DMTF Redfish API for out-of-band management, an onboard TPM 2.0 secures the platform, and an optional Microchip CEC1736 Hardware Root of Trust is available. Three PCIe 5.0 x16 slots — one double-wide for a display GPU and two single-wide with x8 signals — allow an NVIDIA RTX PRO Blackwell graphics card or additional NICs.\u003c\/p\u003e\n\u003ch3\u003e1600 W 80 PLUS Platinum Power in a 245.7 × 529 × 570.4 mm Tower\u003c\/h3\u003e\n\u003cp\u003eA single 1600 W ATX 80 PLUS Platinum power supply (20 A circuit required) feeds the system. The chassis measures 245.7 mm wide, 529.0 mm high and 570.4 mm deep, rated for 10–30 °C operation, and carries CE and FCC Class A certification alongside NVIDIA and Ubuntu 24.04 LTS certification.\u003c\/p\u003e\n\u003ch3\u003eAvailable in Singapore Through SourceIT\u003c\/h3\u003e\n\u003cp\u003eSourceIT supplies the MSI XpertStation WS300 with a full GST invoice, local warranty support and enterprise procurement assistance. MSI does not publish a fixed warranty term for the WS300; SourceIT confirms the applicable MSI enterprise support term in writing at quotation. For desktop-scale AI, see the NVIDIA DGX Spark; for alternative GB300 stations, see the HP ZGX Fury G1n, ASUS ExpertCenter Pro ET900N G3 and GIGABYTE W775-V10-L01.\u003c\/p\u003e\n\u003cdetails class=\"pd-acc\"\u003e\u003csummary style=\"cursor: pointer; font-weight: 700;\"\u003eMSI XpertStation WS300 Datasheet — Technical Specifications\u003c\/summary\u003e\u003cdiv class=\"pd-acc-body\"\u003e\n\u003ch3\u003eGeneral\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eProduct Name\u003c\/td\u003e\n\u003ctd\u003eMSI XpertStation WS300\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eModel \/ Part Number\u003c\/td\u003e\n\u003ctd\u003eWS300T60L (MSI product code CT60-S8060)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eCategory\u003c\/td\u003e\n\u003ctd\u003eEnterprise AI workstation \/ deskside AI supercomputer (NVIDIA DGX Station GB300 platform)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eForm Factor\u003c\/td\u003e\n\u003ctd\u003eTower (NVIDIA DGX Station)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eManufacturer Support\u003c\/td\u003e\n\u003ctd\u003eMSI enterprise support — term not published by MSI for this model; confirmed at quotation\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eLocal Warranty (Singapore)\u003c\/td\u003e\n\u003ctd\u003eLocal warranty support, procurement and GST invoice by SourceIT\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\u003ch3\u003eProcessor \u0026amp; GPU\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eCompute Platform\u003c\/td\u003e\n\u003ctd\u003eNVIDIA DGX Station architecture — NVIDIA GB300 Grace Blackwell Ultra Desktop Superchip\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eCPU\u003c\/td\u003e\n\u003ctd\u003e1 × NVIDIA Grace CPU, 72 Arm Neoverse V2 cores\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eGPU\u003c\/td\u003e\n\u003ctd\u003e1 × NVIDIA Blackwell Ultra GPU (fifth-generation Tensor Cores, second-generation Transformer Engine)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eCPU–GPU Interconnect\u003c\/td\u003e\n\u003ctd\u003eNVIDIA NVLink-C2C, 900 GB\/s (per NVIDIA DGX Station datasheet)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eAI Performance\u003c\/td\u003e\n\u003ctd\u003eUp to 20 PFLOPS FP4 Tensor Core (with sparsity), per NVIDIA — 10 PFLOPS FP8\/FP6, 5 PFLOPS FP16\/BF16\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eModel Support\u003c\/td\u003e\n\u003ctd\u003eAI models up to 1 trillion parameters, per NVIDIA\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eGPU Partitioning\u003c\/td\u003e\n\u003ctd\u003eMulti-Instance GPU (MIG) — up to 7 instances\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\u003ch3\u003eMemory\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eTotal Coherent Memory\u003c\/td\u003e\n\u003ctd\u003e748 GB unified CPU + GPU memory space\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eGPU Memory\u003c\/td\u003e\n\u003ctd\u003e252 GB HBM3e, 7.1 TB\/s\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eCPU Memory\u003c\/td\u003e\n\u003ctd\u003e496 GB LPDDR5X, 396 GB\/s\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\u003ch3\u003eStorage\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eBoot Storage\u003c\/td\u003e\n\u003ctd\u003e2 × M.2 2280 PCIe 5.0 x4 NVMe (from CPU), shipped as 2 TB RAID 1\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eExpansion Storage\u003c\/td\u003e\n\u003ctd\u003e2 × M.2 2280 PCIe 6.0 x4 NVMe (from ConnectX-8), open for expansion\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eWireless Slot\u003c\/td\u003e\n\u003ctd\u003e1 × M.2 2230 PCIe 2.0 x1 for Wi-Fi \/ Bluetooth (optional module)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\u003ch3\u003eExpansion\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003ePCIe Slots\u003c\/td\u003e\n\u003ctd\u003e1 × PCIe 5.0 x16 FHFL double-wide (display GPU), 2 × PCIe 5.0 x16 FHFL single-wide (x8 signals)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\u003ch3\u003eNetworking\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eHigh-Speed Networking\u003c\/td\u003e\n\u003ctd\u003e2 × QSFP112 400G Ethernet ports, NVIDIA ConnectX-8 SuperNIC (up to 800 Gb\/s)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eGeneral Ethernet\u003c\/td\u003e\n\u003ctd\u003e1 × 10GBase-T, Marvell AQC113\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eManagement LAN\u003c\/td\u003e\n\u003ctd\u003e1 × 1000Base-T dedicated server management port\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\u003ch3\u003eManagement \u0026amp; Security\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eBMC\u003c\/td\u003e\n\u003ctd\u003eASPEED AST2600 with AMI MegaRAC-based firmware; IPMI 2.0 and DMTF Redfish API\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eTrusted Platform Module\u003c\/td\u003e\n\u003ctd\u003eOnboard TPM 2.0\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eHardware Root of Trust\u003c\/td\u003e\n\u003ctd\u003eOptional Microchip CEC1736\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\u003ch3\u003eOperating System \u0026amp; Software\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eOperating System\u003c\/td\u003e\n\u003ctd\u003eUbuntu 24.04 LTS with NVIDIA AI Developer Tools preinstalled\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eCertifications\u003c\/td\u003e\n\u003ctd\u003eNVIDIA certified; Ubuntu certified\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\u003ch3\u003ePower, Cooling, Physical \u0026amp; Environmental\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003ePower Supply\u003c\/td\u003e\n\u003ctd\u003e1 × 1600 W ATX 80 PLUS Platinum; 20 A circuit required\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eCooling\u003c\/td\u003e\n\u003ctd\u003eLiquid cooling module for up to 1400 W CPU + GPU; 1 × 120 × 25 mm system fan\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eDimensions (W × H × D)\u003c\/td\u003e\n\u003ctd\u003e245.7 × 529.0 × 570.4 mm (9.67 × 20.83 × 22.46 in)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eWeight\u003c\/td\u003e\n\u003ctd\u003eNot published by MSI\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eOperating Temperature\u003c\/td\u003e\n\u003ctd\u003e10 – 30 °C, 10 – 80 % humidity; non-operating 5 – 40 °C\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eRegulatory\u003c\/td\u003e\n\u003ctd\u003eCE, FCC Class A\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\u003c\/div\u003e\u003c\/details\u003e\n\u003cdetails class=\"pd-acc\"\u003e\u003csummary style=\"cursor: pointer; font-weight: 700;\"\u003eCompatibility \u0026amp; Software Support\u003c\/summary\u003e\u003cdiv class=\"pd-acc-body\"\u003e\n\u003ch3\u003eNVIDIA AI Software Stack\u003c\/h3\u003e\n\u003cp\u003eShips with Ubuntu Linux and the NVIDIA AI Developer Tools preinstalled — the same NVIDIA software foundation used across the DGX family. CUDA, cuDNN, TensorRT and TensorRT-LLM, NVIDIA NeMo, NVIDIA NIM microservices, Triton Inference Server, RAPIDS and NGC containers run natively on the Grace Blackwell Ultra platform, alongside PyTorch, TensorFlow, JAX, vLLM, Hugging Face Transformers, DeepSpeed and ONNX Runtime. Docker and Kubernetes deployment are supported for containerised MLOps pipelines. NVIDIA AI Enterprise is available as a separate subscription for supported production deployments.\u003c\/p\u003e\n\u003ch3\u003eArm64 Architecture — Plan Your Software Accordingly\u003c\/h3\u003e\n\u003cp\u003eThe Grace CPU is an Arm Neoverse V2 processor, so the operating system, drivers and container images are arm64 (aarch64) builds. Mainstream AI frameworks and NVIDIA NGC containers are published for arm64, but proprietary x86-only applications will not run natively — check any commercial software you rely on before ordering. NVIDIA has announced a separate DGX Station for Windows SKU (Windows plus Windows Subsystem for Linux) for a later release; the units SourceIT supplies today ship with Ubuntu.\u003c\/p\u003e\n\u003ch3\u003eClustering and Networking Compatibility\u003c\/h3\u003e\n\u003cp\u003eThe two NVIDIA ConnectX-8 SuperNIC QSFP112 ports run at 400 Gb\/s each (up to 800 Gb\/s aggregate) with RDMA over Converged Ethernet, GPUDirect RDMA and GPUDirect Storage support, so two or more systems can be linked for distributed training, or connected to an NVMe-over-fabric storage array. QSFP112 transceivers or direct-attach cables are ordered separately to match your switch fabric — SourceIT will quote compatible optics alongside the system. The 10GbE RJ45 port covers everyday site networking, and the dedicated management LAN keeps out-of-band BMC traffic isolated from production data.\u003c\/p\u003e\n\u003ch3\u003eMSI Enterprise Platform Compatibility\u003c\/h3\u003e\n\u003cp\u003eThe WS300 is listed on MSI's Enterprise Platform Solutions (EPS) catalogue as WS300T60L and is NVIDIA-certified and Ubuntu 24.04 LTS-certified. Its ASPEED AST2600 BMC exposes IPMI 2.0 and the DMTF Redfish API, so it integrates with standard data-centre management tooling alongside MSI's rack servers and other Redfish-capable infrastructure.\u003c\/p\u003e\n\u003c\/div\u003e\u003c\/details\u003e\n\u003cdetails class=\"pd-acc\"\u003e\u003csummary style=\"cursor: pointer; font-weight: 700;\"\u003ePackage Contents \u0026amp; Ordering Information\u003c\/summary\u003e\u003cdiv class=\"pd-acc-body\"\u003e\n\u003ch3\u003eWhat Ships With the System\u003c\/h3\u003e\n\u003cul\u003e\n\u003cli\u003eMSI XpertStation WS300 (WS300T60L) tower with NVIDIA GB300 Grace Blackwell Ultra Desktop Superchip\u003c\/li\u003e\n\u003cli\u003e2 TB NVMe M.2 boot storage configured as RAID 1\u003c\/li\u003e\n\u003cli\u003e1600 W 80 PLUS Platinum power supply (installed) and power cord\u003c\/li\u003e\n\u003cli\u003eMSI quick installation guide and enterprise support documentation\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003cp\u003eQSFP112 transceivers or DAC cables, additional PCIe 6.0 NVMe SSDs, the optional Wi-Fi \/ Bluetooth module and an optional NVIDIA RTX PRO Blackwell display GPU are quoted separately.\u003c\/p\u003e\n\u003ch3\u003eNVIDIA DGX Station GB300 Systems at SourceIT Singapore\u003c\/h3\u003e\n\u003cp\u003eThe MSI XpertStation WS300 is in stock at SourceIT Singapore and quoted with GST invoice; enterprise support terms are confirmed in writing at quotation.\u003c\/p\u003e\n\u003cul\u003e\n\u003cli\u003e\n\u003ca href=\"\/products\/asus-expertcenter-pro-et900n-g3-gb300-ai-supercomputer\"\u003eASUS ExpertCenter Pro ET900N G3\u003c\/a\u003e — NVIDIA DGX Station GB300 platform, same 748 GB \/ 20 PFLOPS core compute\u003c\/li\u003e\n\u003cli\u003e\n\u003ca href=\"\/products\/hp-zgx-fury-g1n-gb300-ai-workstation-with-nvidia-blackwell-ultra\"\u003eHP ZGX Fury AI Station G1n\u003c\/a\u003e — NVIDIA DGX Station GB300 platform, same 748 GB \/ 20 PFLOPS core compute\u003c\/li\u003e\n\u003cli\u003e\n\u003ca href=\"\/products\/gigabyte-w775-v10-l01-deskside-ai-supercomputer-with-nvidia-gb300-blackwell\"\u003eGIGABYTE W775-V10-L01\u003c\/a\u003e — NVIDIA DGX Station GB300 platform, same 748 GB \/ 20 PFLOPS core compute\u003c\/li\u003e\n\u003cli\u003e\n\u003ca href=\"\/products\/nvidia-dgx-spark-ai-supercomputer-4tb-940-54242-0007-000\"\u003eNVIDIA DGX Spark\u003c\/a\u003e — desktop-scale GB10 system (128 GB unified memory, 1 PFLOP FP4) for individual developers\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003cp\u003eBrowse the full \u003ca href=\"\/collections\/nvidia-dgx-spark\"\u003eNVIDIA DGX Spark \u0026amp; DGX Station collection\u003c\/a\u003e. For volume, project, research-grant or education pricing, GST invoicing and current lead times, email \u003ca href=\"mailto:sales@sourceit.com.sg\"\u003esales@sourceit.com.sg\u003c\/a\u003e.\u003c\/p\u003e\n\u003c\/div\u003e\u003c\/details\u003e\n\u003cdetails class=\"pd-acc\"\u003e\u003csummary style=\"cursor: pointer; font-weight: 700;\"\u003eDatasheet \u0026amp; Downloads\u003c\/summary\u003e\u003cdiv class=\"pd-acc-body\"\u003e\n\u003cp\u003eOfficial MSI documentation for the XpertStation WS300:\u003c\/p\u003e\n\u003cul\u003e\n\u003cli\u003e\u003ca href=\"https:\/\/download-2.msi.com\/archive\/mnu_exe\/server\/WS300-D08262026.pdf\" target=\"_blank\" rel=\"noopener\"\u003eMSI XpertStation WS300 specification sheet (PDF)\u003c\/a\u003e\u003c\/li\u003e\n\u003cli\u003e\u003ca href=\"https:\/\/eps.msi.com\/en\/product\/barebones\/WS300T60L\" target=\"_blank\" rel=\"noopener\"\u003eMSI EPS product page — WS300T60L\u003c\/a\u003e\u003c\/li\u003e\n\u003cli\u003e\u003ca href=\"https:\/\/www.msi.com\/Landing\/NVIDIA-DGX-STATION\" target=\"_blank\" rel=\"noopener\"\u003eMSI NVIDIA DGX Station overview\u003c\/a\u003e\u003c\/li\u003e\n\u003cli\u003e\u003ca href=\"https:\/\/www.nvidia.com\/en-sg\/products\/workstations\/dgx-station\/\" target=\"_blank\" rel=\"noopener\"\u003eNVIDIA DGX Station (GB300) — official platform page (Singapore)\u003c\/a\u003e\u003c\/li\u003e\n\u003cli\u003e\u003ca href=\"https:\/\/www.nvidia.com\/content\/dam\/en-zz\/Solutions\/products\/workstations\/dgx-station\/workstation-datasheet-dgx-station-nvidia-us-5543600-web.pdf\" target=\"_blank\" rel=\"noopener\"\u003eNVIDIA DGX Station datasheet (PDF)\u003c\/a\u003e\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003c\/div\u003e\u003c\/details\u003e\n\u003cdetails class=\"pd-acc\"\u003e\u003csummary style=\"cursor: pointer; font-weight: 700;\"\u003eFrequently Asked Questions — MSI XpertStation WS300\u003c\/summary\u003e\u003cdiv class=\"pd-acc-body\"\u003e\n\u003ch3\u003eWhat is the MSI XpertStation WS300?\u003c\/h3\u003e\n\u003cp\u003eMSI's enterprise AI workstation built on the NVIDIA DGX Station GB300 platform: a 72-core Grace CPU and Blackwell Ultra GPU sharing 748 GB of coherent memory, up to 20 PFLOPS FP4, liquid cooled, in a deskside tower — for training, fine-tuning and serving AI models of up to 1 trillion parameters on premises.\u003c\/p\u003e\n\u003ch3\u003eHow does the WS300 compare to the NVIDIA DGX Spark?\u003c\/h3\u003e\n\u003cp\u003eDifferent class entirely: the WS300 delivers up to 20 PFLOPS FP4 and 748 GB coherent memory for models up to 1 trillion parameters, versus the DGX Spark's 1 PFLOP and 128 GB (about 200 billion parameters). Choose the Spark for individual developers; the WS300 for teams and production-scale AI.\u003c\/p\u003e\n\u003ch3\u003eWhat storage does it ship with?\u003c\/h3\u003e\n\u003cp\u003eA 2 TB NVMe RAID 1 boot pair on the two PCIe 5.0 M.2 slots, with two PCIe 6.0 M.2 slots free for dataset and checkpoint storage.\u003c\/p\u003e\n\u003ch3\u003eIs the WS300 liquid cooled?\u003c\/h3\u003e\n\u003cp\u003eYes. MSI fits a liquid cooling module rated for up to 1400 W of CPU + GPU load, supported by a 120 mm system fan, inside the tower — no external liquid-cooled rack is required.\u003c\/p\u003e\n\u003ch3\u003eCan multiple team members share one WS300?\u003c\/h3\u003e\n\u003cp\u003eYes — Multi-Instance GPU (MIG) partitions the Blackwell Ultra GPU into up to seven isolated instances, so several users or workloads run concurrently with guaranteed resources.\u003c\/p\u003e\n\u003ch3\u003eDoes it need a server room?\u003c\/h3\u003e\n\u003cp\u003eNo rack is needed, but plan an air-conditioned space within MSI's 10–30 °C operating range and a dedicated 20 A power circuit for the 1600 W Platinum power supply.\u003c\/p\u003e\n\u003ch3\u003eCan I add a display graphics card?\u003c\/h3\u003e\n\u003cp\u003eYes. The double-wide PCIe 5.0 x16 slot is intended for a display GPU such as an NVIDIA RTX PRO Blackwell card, and two further x16 slots (x8 signals) take additional NICs or accelerators — quoted separately.\u003c\/p\u003e\n\u003ch3\u003eWhat warranty does MSI provide?\u003c\/h3\u003e\n\u003cp\u003eMSI does not publish a fixed warranty term for the WS300 on its product pages. SourceIT confirms the applicable MSI enterprise support term for Singapore in writing at quotation and provides local warranty support.\u003c\/p\u003e\n\u003ch3\u003eHow do I purchase the WS300 in Singapore?\u003c\/h3\u003e\n\u003cp\u003eSourceIT is an authorised IT reseller in Singapore and supplies the MSI XpertStation WS300 with GST invoice, local warranty support and enterprise procurement assistance — contact sales@sourceit.com.sg for configuration options, volume pricing and lead times.\u003c\/p\u003e\n\u003c\/div\u003e\u003c\/details\u003e\n\u003ch3\u003eProduct Video\u003c\/h3\u003e\n\u003cp\u003e\u003ciframe width=\"560\" height=\"315\" src=\"https:\/\/www.youtube.com\/embed\/LaTQE00kCvA\" title=\"NVIDIA DGX Station XpertStation WS300 | MSI\" frameborder=\"0\" allow=\"accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture\" allowfullscreen\u003e\u003c\/iframe\u003e\u003c\/p\u003e\n\u003c\/div\u003e","brand":"Nvidia","offers":[{"title":"Default Title","offer_id":51421413441700,"sku":"XpertStation WS300","price":150000.0,"currency_code":"SGD","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0605\/0136\/0804\/files\/msi-xpertstation-ws300-angle-square.png?v=1790257473"},{"product_id":"hp-zgx-fury-g1n-gb300-ai-workstation-with-nvidia-blackwell-ultra","title":"HP ZGX Fury G1n NVIDIA DGX Station GB300 AI Workstation","description":"\u003cdiv\u003e\n\u003cstyle\u003e\n.pd-acc { border: 1px solid #e2e2e2; border-radius: 8px; margin: 10px 0; overflow: hidden; }\n.pd-acc summary { cursor: pointer; padding: 14px 16px; font-weight: 700; font-size: 1.05em; list-style: none; display: flex; justify-content: space-between; align-items: center; background: #f7f7f7; }\n.pd-acc[open] \u003e summary { border-bottom: 1px solid #e2e2e2; }\n.pd-acc summary::-webkit-details-marker { display: none; }\n.pd-acc summary::after { content: \"+\"; font-size: 1.4em; font-weight: 400; line-height: 1; margin-left: 12px; }\n.pd-acc[open] \u003e summary::after { content: \"\\2212\"; }\n.pd-acc .pd-acc-body { padding: 8px 16px 14px; }\n.pd-spec, .pd-spec tr, .pd-spec td { border: none !important; background: transparent !important; }\n.pd-spec { width: 100%; border-collapse: collapse !important; margin: 6px 0 18px; }\n.pd-spec td { padding: 7px 10px 7px 0 !important; border-bottom: 1px solid #ececec !important; vertical-align: top; font-size: 0.95em; text-align: left; }\n.pd-spec td:first-child { width: 38%; font-weight: 600; color: #333; }\n.pd-spec tr:last-child td { border-bottom: none !important; }\n\u003c\/style\u003e\n\u003ch2\u003eHP ZGX Fury AI Station G1n NVIDIA DGX Station GB300 Grace Blackwell Ultra AI Workstation, 72-Core Grace CPU, 748GB Coherent Memory, 20 PFLOPS FP4, HP Z Turbo Drive PCIe Gen5 NVMe, Dual 400G ConnectX-8, Liquid-Cooled Rackable Tower (E6ZS0PT \/ E6ZS5PT) - 3 Years Local Warranty\u003c\/h2\u003e\n\u003ch3\u003eHP ZGX Fury AI Station G1n — NVIDIA DGX Station GB300 AI Workstation with 3-Year Onsite Warranty\u003c\/h3\u003e\n\u003cp\u003eThe HP ZGX Fury AI Station G1n is HP's build of the NVIDIA DGX Station GB300 platform: an NVIDIA GB300 Grace Blackwell Ultra Desktop Superchip pairing a 72-core Grace CPU with a Blackwell Ultra GPU across 748 GB of coherent memory, delivering up to 20 petaFLOPS of FP4 AI compute for training, fine-tuning and deploying LLMs of up to 1 trillion parameters, AI agents, generative AI and scientific simulation — entirely within your own infrastructure. HP positions it for development and fine-tuning of 100-billion-class models and inference on trillion-parameter-class models from a single deskside or rack-mounted system.\u003c\/p\u003e\n\u003cp\u003eUnlike other GB300 stations, the ZGX Fury G1n is backed in Singapore by HP's 3-year (3\/3\/3) limited warranty covering parts, labour and onsite repair, and it is available in factory-built configurations with an NVIDIA RTX PRO Blackwell display GPU and up to 12 TB of HP Z Turbo Drive PCIe Gen5 NVMe storage. It is the natural choice for enterprises, financial institutions, government agencies and research institutes in Singapore that already standardise on HP Z workstations and want data-centre-class AI with a familiar support model.\u003c\/p\u003e\n\u003ch3\u003eNVIDIA GB300 Grace Blackwell Ultra Desktop Superchip — 20 PFLOPS of FP4 AI Compute\u003c\/h3\u003e\n\u003cp\u003eThe GB300 Desktop Superchip joins one NVIDIA Grace CPU with 72 Arm Neoverse V2 cores to one NVIDIA Blackwell Ultra GPU over a 900 GB\/s NVLink-C2C interconnect, so the CPU and GPU address one coherent memory space rather than shuttling data across PCIe. Per NVIDIA's DGX Station datasheet the Blackwell Ultra GPU delivers up to 20 PFLOPS of FP4 Tensor Core performance with sparsity, 10 PFLOPS FP8\/FP6, 5 PFLOPS FP16\/BF16, 2.5 PFLOPS TF32 and 80 TFLOPS FP32, with seven NVDEC and seven nvJPEG decoders for video and image pipelines. Multi-Instance GPU (MIG) partitions the GPU into up to seven isolated instances so a team can share one system for concurrent training, fine-tuning and inference jobs.\u003c\/p\u003e\n\u003ch3\u003e748 GB Coherent Unified Memory for Frontier-Scale Models\u003c\/h3\u003e\n\u003cp\u003e252 GB of HBM3e GPU memory at 7.1 TB\/s and 496 GB of LPDDR5X CPU memory (four 128 GB SOCAMM LPDDR5X-6400 modules) at 396 GB\/s unite over the 900 GB\/s NVLink-C2C link into a single 748 GB coherent memory space — enough to load trillion-parameter models at FP4, run long-context RAG pipelines, fine-tune foundation models and drive digital-twin and engineering simulation without CPU–GPU offloading. NVIDIA notes that two linked DGX Station systems present more than 1.4 TB of coherent memory and up to 40 petaFLOPS of AI compute.\u003c\/p\u003e\n\u003ch3\u003eFactory Configurations — RTX PRO 2000 or RTX PRO 4000 Blackwell, 8 TB or 12 TB NVMe\u003c\/h3\u003e\n\u003cp\u003eEvery ZGX Fury G1n shares the same GB300 Superchip, 748 GB coherent memory, 72-core Grace CPU, dual 400G ConnectX-8 networking and 3\/3\/3 warranty; the configurations differ in the professional display GPU and NVMe storage. E6ZS0PT#AB4 pairs an NVIDIA RTX PRO 2000 Blackwell (16 GB GDDR7) with 4 × 2 TB HP Z Turbo Drive PCIe Gen5 NVMe SSDs (8 TB). E6ZS5PT#AB4 steps up to an NVIDIA RTX PRO 4000 Blackwell SFF (24 GB GDDR7) with 2 × 2 TB + 2 × 4 TB NVMe (12 TB). A headless configuration with no discrete graphics card and 4 × 2 TB self-encrypting drives (SED) is available to order for remote-accessed or compliance-driven deployments — part number confirmed at quotation. Select your configuration from the dropdown above.\u003c\/p\u003e\n\u003ch3\u003eDual 400G ConnectX-8 Networking, Gen5 \/ Gen6 NVMe and PCIe 5.0 Expansion\u003c\/h3\u003e\n\u003cp\u003eTwo QSFP112 ports on the NVIDIA ConnectX-8 SuperNIC run at 400 Gb\/s each (up to 800 Gb\/s) for linking two systems, private AI clusters and NVMe-over-fabric storage, alongside a 10 GbE RJ45 port for site networking and a dedicated 1 GbE BMC port for out-of-band management. Two embedded M.2 PCIe Gen5 slots off the CPU hold the OS in software RAID 1, and two M.2 PCIe Gen6 slots off the ConnectX-8 hold data in software RAID 0 — all SED-capable. Three PCIe slots (one Gen5 x16 and two Gen5 x16 at x8 electrical) take the display GPU and further NICs or accelerators. Front I\/O provides two USB-C 10 Gbps, two USB-A 5 Gbps and an audio jack; the rear adds four USB-A 10 Gbps, audio in\/out and BMC-only micro-USB and Mini DisplayPort 2.0.\u003c\/p\u003e\n\u003ch3\u003eLiquid-Cooled Rackable Tower — Deskside Today, 5U Rack Tomorrow\u003c\/h3\u003e\n\u003cp\u003eHP builds the ZGX Fury G1n as a liquid-cooled rackable tower — 44.6 × 22.15 × 66.7 cm in standard tower orientation, starting at 41 kg — with front-to-back airflow and rack rails included, so it can sit beside a desk today and move into a standard 5U rack position later without a chassis change. HP specifies a dedicated 200–240 V AC circuit with a C20 inlet for the 1,600 W-class DGX Station platform; it is not intended for a standard 120 V outlet. NVIDIA-approved partner-developed BIOS and BMC firmware provide secure remote management for continuous 24\/7 operation.\u003c\/p\u003e\n\u003ch3\u003eUbuntu 24.04 LTS, NVIDIA AI Developer Tools, HP Z Runtime and HP Z Toolkit\u003c\/h3\u003e\n\u003cp\u003eThe ZGX Fury ships with Ubuntu 24.04 LTS and the NVIDIA AI Developer Tools — drivers, CUDA and container tooling — preinstalled, the same NVIDIA software foundation used across the DGX family, so TensorRT-LLM, NVIDIA NeMo, NIM microservices, Triton, RAPIDS and NGC containers run natively alongside PyTorch, JAX, vLLM and Hugging Face Transformers. HP adds HP Z Runtime, a preinstalled command-line tool for pulling, serving and managing models, and the HP Z Toolkit, a free Visual Studio Code extension for an x86 Windows 11 or Ubuntu 24.04 client that handles discovery, pairing, remote access and SSH management. HP states the system is certified for Red Hat Enterprise Linux and that Windows support is coming; the units SourceIT supplies today ship with Ubuntu. NVIDIA AI Enterprise is available as a separate subscription.\u003c\/p\u003e\n\u003ch3\u003eAvailable in Singapore Through SourceIT\u003c\/h3\u003e\n\u003cp\u003eSourceIT is an authorised IT reseller in Singapore and supplies the HP ZGX Fury AI Station G1n with a full GST invoice, HP's 3-year onsite warranty and enterprise procurement assistance — email \u003ca href=\"mailto:sales@sourceit.com.sg\"\u003esales@sourceit.com.sg\u003c\/a\u003e or use the Request for Quotation button for configuration options, lead times and pricing. For desktop-scale AI, see the \u003ca href=\"\/products\/nvidia-dgx-spark-ai-supercomputer-4tb-940-54242-0007-000\"\u003eNVIDIA DGX Spark\u003c\/a\u003e or the \u003ca href=\"\/products\/hp-zgx-nano-g1n-ai-station-4tb\"\u003eHP ZGX Nano G1n\u003c\/a\u003e; for alternative GB300 stations, see the \u003ca href=\"\/products\/msi-xpertstation-ws300-nvidia-dgx-station-ai-supercomputer\"\u003eMSI XpertStation WS300\u003c\/a\u003e, \u003ca href=\"\/products\/asus-expertcenter-pro-et900n-g3-gb300-ai-supercomputer\"\u003eASUS ExpertCenter Pro ET900N G3\u003c\/a\u003e, \u003ca href=\"\/products\/gigabyte-w775-v10-l01-deskside-ai-supercomputer-with-nvidia-gb300-blackwell\"\u003eGIGABYTE W775-V10-L01\u003c\/a\u003e and \u003ca href=\"\/products\/supermicro-super-ai-station-ars-511gd-nb-lcc-nvidia-dgx-station-gb300\"\u003eSupermicro ARS-511GD-NB-LCC\u003c\/a\u003e in the \u003ca href=\"\/collections\/nvidia-dgx-spark\"\u003eNVIDIA DGX Spark \u0026amp; DGX Station collection\u003c\/a\u003e.\u003c\/p\u003e\n\u003cdetails class=\"pd-acc\"\u003e\u003csummary style=\"cursor: pointer; font-weight: 700;\"\u003eHP ZGX Fury AI Station G1n Datasheet — Technical Specifications\u003c\/summary\u003e\u003cdiv class=\"pd-acc-body\"\u003e\n\u003ch3\u003eGeneral\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eProduct Name\u003c\/td\u003e\n\u003ctd\u003eHP ZGX Fury AI Station G1n\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003ePart Numbers (Singapore)\u003c\/td\u003e\n\u003ctd\u003eE6ZS0PT#AB4 (RTX PRO 2000 Blackwell, 4 × 2 TB); E6ZS5PT#AB4 (RTX PRO 4000 Blackwell SFF, 2 × 2 TB + 2 × 4 TB); no-discrete-GPU SED configuration to order\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eCategory\u003c\/td\u003e\n\u003ctd\u003eEnterprise AI workstation \/ deskside AI supercomputer (NVIDIA DGX Station GB300)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eForm Factor\u003c\/td\u003e\n\u003ctd\u003eLiquid-cooled rackable tower; 5U rack with included rails\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eWarranty\u003c\/td\u003e\n\u003ctd\u003eHP 3-year (3\/3\/3) limited warranty — 3 years parts, labour and onsite repair\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eLocal Warranty (Singapore)\u003c\/td\u003e\n\u003ctd\u003e3 Years HP onsite; procurement and GST invoice by SourceIT\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\n\u003ch3\u003eProcessor \u0026amp; GPU\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eCompute Platform\u003c\/td\u003e\n\u003ctd\u003eNVIDIA DGX Station architecture — NVIDIA GB300 Grace Blackwell Ultra Desktop Superchip\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eCPU\u003c\/td\u003e\n\u003ctd\u003e1 × NVIDIA Grace CPU, 72 Arm Neoverse V2 cores, soldered\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eGPU\u003c\/td\u003e\n\u003ctd\u003e1 × NVIDIA Blackwell Ultra GPU\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eCPU–GPU Interconnect\u003c\/td\u003e\n\u003ctd\u003eNVIDIA NVLink-C2C, 900 GB\/s (per NVIDIA DGX Station datasheet)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eGPU Partitioning\u003c\/td\u003e\n\u003ctd\u003eMulti-Instance GPU (MIG) — up to 7 instances\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eProfessional Display Graphics\u003c\/td\u003e\n\u003ctd\u003eE6ZS0PT: NVIDIA RTX PRO 2000 Blackwell (16 GB GDDR7); E6ZS5PT: NVIDIA RTX PRO 4000 Blackwell SFF (24 GB GDDR7); headless configuration: none\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\n\u003ch3\u003eAI Performance (per NVIDIA DGX Station datasheet)\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eFP4 Tensor Core\u003c\/td\u003e\n\u003ctd\u003eUp to 20 PFLOPS (with sparsity)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eFP8 \/ FP6 Tensor Core\u003c\/td\u003e\n\u003ctd\u003e10 PFLOPS\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eFP16 \/ BF16 Tensor Core\u003c\/td\u003e\n\u003ctd\u003e5 PFLOPS\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eTF32 Tensor Core \/ FP32 \/ FP64\u003c\/td\u003e\n\u003ctd\u003e2.5 PFLOPS \/ 80 TFLOPS \/ 1.3 TFLOPS\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eModel Support\u003c\/td\u003e\n\u003ctd\u003eAI models up to 1 trillion parameters (FP4), per NVIDIA and HP\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eDecoders\u003c\/td\u003e\n\u003ctd\u003e7 × NVDEC, 7 × nvJPEG\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\n\u003ch3\u003eMemory\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eTotal Coherent Memory\u003c\/td\u003e\n\u003ctd\u003e748 GB unified CPU + GPU memory space\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eGPU Memory\u003c\/td\u003e\n\u003ctd\u003e252 GB HBM3e, 7.1 TB\/s\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eCPU Memory\u003c\/td\u003e\n\u003ctd\u003e496 GB LPDDR5X-6400, 396 GB\/s\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eCPU Memory Modules\u003c\/td\u003e\n\u003ctd\u003e4 × 128 GB SOCAMM LPDDR5X (onboard)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\n\u003ch3\u003eStorage\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eOS Storage\u003c\/td\u003e\n\u003ctd\u003e2 × embedded M.2 PCIe Gen5 (from CPU), software RAID 1 through host OS\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eData Storage\u003c\/td\u003e\n\u003ctd\u003e2 × embedded M.2 PCIe Gen6 (from ConnectX-8), software RAID 0 through host OS\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eInstalled Drives\u003c\/td\u003e\n\u003ctd\u003eE6ZS0PT: 4 × 2 TB HP Z Turbo Drive PCIe Gen5 M.2 NVMe (8 TB); E6ZS5PT: 2 × 2 TB + 2 × 4 TB (12 TB); headless configuration: 4 × 2 TB self-encrypting NVMe\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eEncryption\u003c\/td\u003e\n\u003ctd\u003eSED capable\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\n\u003ch3\u003eNetworking\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eHigh-Speed Networking\u003c\/td\u003e\n\u003ctd\u003e2 × QSFP112, 400 Gb\/s per port, NVIDIA ConnectX-8 SuperNIC (up to 800 Gb\/s, Ethernet)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eGeneral Ethernet\u003c\/td\u003e\n\u003ctd\u003e1 × RJ45 10 GbE\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eManagement LAN\u003c\/td\u003e\n\u003ctd\u003e1 × RJ45 1 GbE (BMC, system management only)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\n\u003ch3\u003eI\/O, Expansion \u0026amp; Display\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eFront I\/O\u003c\/td\u003e\n\u003ctd\u003e2 × USB Type-C 10 Gbps, 2 × USB Type-A 5 Gbps, 1 × universal audio jack\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eRear I\/O\u003c\/td\u003e\n\u003ctd\u003e4 × USB Type-A 10 Gbps, microphone, audio-in, line-out; 1 × micro USB (BMC); 1 × Mini DisplayPort 2.0 (BMC systems management only); Kensington lock slot\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003ePCIe Slots\u003c\/td\u003e\n\u003ctd\u003e3 — 1 × PCIe Gen5 x16; 2 × PCIe Gen5 x16 (x8 electrical)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\n\u003ch3\u003eSoftware, Management \u0026amp; Security\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eOperating System\u003c\/td\u003e\n\u003ctd\u003eUbuntu 24.04 LTS with NVIDIA AI Developer Tools (arm64); Red Hat Enterprise Linux certified per HP\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eHP Software\u003c\/td\u003e\n\u003ctd\u003eHP Z Runtime (model CLI, preinstalled); HP Z Toolkit (VS Code extension for Windows 11 \/ Ubuntu 24.04 x86 clients)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eNVIDIA AI Stack\u003c\/td\u003e\n\u003ctd\u003eCUDA, cuDNN, TensorRT \/ TensorRT-LLM, NVIDIA NIM, NVIDIA NeMo, Triton, RAPIDS, NGC containers; NVIDIA AI Enterprise by separate subscription\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eBIOS \/ BMC\u003c\/td\u003e\n\u003ctd\u003eNVIDIA-approved partner-developed BIOS and BMC firmware; secure remote management for continuous 24\/7 operation\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\n\u003ch3\u003ePower, Cooling \u0026amp; Physical\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eCooling\u003c\/td\u003e\n\u003ctd\u003eLiquid cooled (integrated in chassis), front intake and rear exhaust\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003ePower\u003c\/td\u003e\n\u003ctd\u003e1,600 W total system power (NVIDIA DGX Station platform); dedicated 200–240 V AC circuit with C20 inlet required per HP, not for 120 V outlets\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eDimensions (H × W × D)\u003c\/td\u003e\n\u003ctd\u003e44.6 × 22.15 × 66.7 cm, standard tower orientation; rack rails included for 5U rack mounting\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eWeight\u003c\/td\u003e\n\u003ctd\u003eStarting at 41 kg\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\n\u003c\/div\u003e\u003c\/details\u003e\n\u003cdetails class=\"pd-acc\"\u003e\u003csummary style=\"cursor: pointer; font-weight: 700;\"\u003eCompatibility \u0026amp; Software Support\u003c\/summary\u003e\u003cdiv class=\"pd-acc-body\"\u003e\n\u003ch3\u003eNVIDIA AI Software Stack\u003c\/h3\u003e\n\u003cp\u003eShips with Ubuntu 24.04 LTS and the NVIDIA AI Developer Tools preinstalled — the same NVIDIA software foundation used across the DGX family. CUDA, TensorRT-LLM, NVIDIA NeMo, NIM microservices, Triton, RAPIDS and NGC containers run natively on the Grace Blackwell Ultra platform, alongside PyTorch, JAX, vLLM, Hugging Face Transformers and ONNX Runtime; Docker and Kubernetes cover containerised MLOps pipelines, and HP's Z Runtime CLI pulls, serves and manages models out of the box.\u003c\/p\u003e\n\u003ch3\u003eArm64 Architecture — Plan Your Software Accordingly\u003c\/h3\u003e\n\u003cp\u003eThe Grace CPU is an Arm Neoverse V2 processor, so the operating system, drivers and container images are arm64 (aarch64) builds. Mainstream AI frameworks and NVIDIA NGC containers are published for arm64, but proprietary x86-only applications will not run natively — check any commercial software you rely on before ordering. HP states that Windows support for the ZGX Fury is coming soon and NVIDIA has announced DGX Station for Windows; the units SourceIT supplies today ship with Ubuntu, and HP lists the system as Red Hat Enterprise Linux certified.\u003c\/p\u003e\n\u003ch3\u003eClustering and Networking Compatibility\u003c\/h3\u003e\n\u003cp\u003eThe two NVIDIA ConnectX-8 SuperNIC QSFP112 ports run at 400 Gb\/s each (up to 800 Gb\/s aggregate), so two systems can be linked for more than 1.4 TB of coherent memory and up to 40 PFLOPS, or connected to an NVMe-over-fabric storage array. QSFP112 transceivers or direct-attach cables are ordered separately to match your fabric — SourceIT quotes compatible optics or DACs with the system; the \u003ca href=\"\/products\/hp-zgx-0-4m-qsfp112-ethernet-dac-cable\"\u003eHP ZGX 0.4 m QSFP112 Ethernet DAC cable (C20MQAA)\u003c\/a\u003e is HP's QSFP112 cable for the ZGX family. The 10 GbE RJ45 port covers site networking, and the dedicated management LAN keeps BMC traffic isolated from production data.\u003c\/p\u003e\n\u003ch3\u003eHP Z Ecosystem\u003c\/h3\u003e\n\u003cp\u003eThe ZGX Fury G1n ships with HP Z Runtime and HP Z Toolkit and includes rack rails so a deployment can grow from deskside to a shared compute room. The optional RTX PRO 2000 \/ 4000 Blackwell display GPUs suit local visualisation, CAD and DCC alongside AI work; the headless SED configuration suits remote-accessed and regulated environments. The same HP Z Toolkit also manages the GB10-based \u003ca href=\"\/products\/hp-zgx-nano-g1n-ai-station-4tb\"\u003eHP ZGX Nano G1n\u003c\/a\u003e, so a team can prototype on the Nano and scale to the Fury inside one toolchain.\u003c\/p\u003e\n\u003c\/div\u003e\u003c\/details\u003e\n\u003cdetails class=\"pd-acc\"\u003e\u003csummary style=\"cursor: pointer; font-weight: 700;\"\u003ePackage Contents \u0026amp; Ordering Information\u003c\/summary\u003e\u003cdiv class=\"pd-acc-body\"\u003e\n\u003ch3\u003eWhat Ships With the System\u003c\/h3\u003e\n\u003cul\u003e\n\u003cli\u003eHP ZGX Fury AI Station G1n with NVIDIA GB300 Grace Blackwell Ultra Desktop Superchip\u003c\/li\u003e\n\u003cli\u003eNVMe storage and display GPU as per the selected configuration (E6ZS0PT#AB4 or E6ZS5PT#AB4; headless SED build to order)\u003c\/li\u003e\n\u003cli\u003eRack rails for 5U rack mounting, power cord and HP documentation\u003c\/li\u003e\n\u003cli\u003eHP 3-year (3\/3\/3) limited warranty with onsite repair\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003cp\u003eQSFP112 transceivers or DAC cables — for example the \u003ca href=\"\/products\/hp-zgx-0-4m-qsfp112-ethernet-dac-cable\"\u003eHP ZGX 0.4 m QSFP112 Ethernet DAC cable\u003c\/a\u003e for linking two units — are quoted separately.\u003c\/p\u003e\n\u003ch3\u003eWhich Configuration?\u003c\/h3\u003e\n\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eE6ZS0PT#AB4 — RTX PRO 2000 Blackwell \/ 4 × 2 TB (this listing, first dropdown option)\u003c\/strong\u003e — balanced build with professional display graphics for local visualisation\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eE6ZS5PT#AB4 — RTX PRO 4000 Blackwell SFF \/ 2 × 2 TB + 2 × 4 TB (this listing, second dropdown option)\u003c\/strong\u003e — more on-board dataset capacity (12 TB) and the stronger visualisation GPU\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eHeadless SED — no graphics card \/ 4 × 2 TB SED (this listing, third dropdown option)\u003c\/strong\u003e — rack-adjacent or remote-accessed deployments, or where self-encrypting drives are mandated by security or compliance policy; part number confirmed at quotation\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003eOther NVIDIA GB10 and DGX Station Systems at SourceIT\u003c\/h3\u003e\n\u003cul\u003e\n\u003cli\u003e\n\u003ca href=\"\/products\/msi-xpertstation-ws300-nvidia-dgx-station-ai-supercomputer\"\u003eMSI XpertStation WS300\u003c\/a\u003e — NVIDIA DGX Station GB300 platform, same 748 GB \/ 20 PFLOPS core compute\u003c\/li\u003e\n\u003cli\u003e\n\u003ca href=\"\/products\/asus-expertcenter-pro-et900n-g3-gb300-ai-supercomputer\"\u003eASUS ExpertCenter Pro ET900N G3\u003c\/a\u003e — NVIDIA DGX Station GB300 platform\u003c\/li\u003e\n\u003cli\u003e\n\u003ca href=\"\/products\/gigabyte-w775-v10-l01-deskside-ai-supercomputer-with-nvidia-gb300-blackwell\"\u003eGIGABYTE W775-V10-L01\u003c\/a\u003e — NVIDIA DGX Station GB300 platform\u003c\/li\u003e\n\u003cli\u003e\n\u003ca href=\"\/products\/supermicro-super-ai-station-ars-511gd-nb-lcc-nvidia-dgx-station-gb300\"\u003eSupermicro Super AI Station ARS-511GD-NB-LCC\u003c\/a\u003e — NVIDIA DGX Station GB300 platform\u003c\/li\u003e\n\u003cli\u003e\n\u003ca href=\"\/products\/hp-zgx-nano-g1n-ai-station-4tb\"\u003eHP ZGX Nano G1n 4TB\u003c\/a\u003e and \u003ca href=\"\/products\/nvidia-dgx-spark-ai-supercomputer-4tb-940-54242-0007-000\"\u003eNVIDIA DGX Spark 4TB\u003c\/a\u003e — desktop-scale GB10 systems (128 GB unified memory, 1 PFLOP FP4); also from \u003ca href=\"\/products\/asus-ascent-gx10-compact-desktop-ai-supercomputer-4tb-gx10-gg0007bn\"\u003eASUS\u003c\/a\u003e, \u003ca href=\"\/products\/dell-pro-max-ai-desktop-pcs-with-nvidia-gb10-blackwell-gpu-4tb\"\u003eDell\u003c\/a\u003e, \u003ca href=\"\/products\/lenovo-thinkstation-pgx-4tb-nvidia-grace-blackwell-30kls00k00\"\u003eLenovo\u003c\/a\u003e, \u003ca href=\"\/products\/gigabyte-ai-top-atom-personal-ai-computer-4tb\"\u003eGIGABYTE\u003c\/a\u003e, \u003ca href=\"\/products\/msi-edgexpert-personal-ai-supercomputer-4tb\"\u003eMSI\u003c\/a\u003e, \u003ca href=\"\/products\/altos-brainsphere%E2%84%A2-gb10-f1-ai-desktop-4tb\"\u003eAcer\u003c\/a\u003e and \u003ca href=\"\/products\/xfusion-fusionxpark-gb10-personal-ai-supercomputer-4tb\"\u003exFusion\u003c\/a\u003e\n\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003cp\u003eBrowse the \u003ca href=\"\/collections\/nvidia-dgx-spark\"\u003eNVIDIA DGX Spark \u0026amp; DGX Station collection\u003c\/a\u003e and the \u003ca href=\"\/collections\/ai-workstations-ai-servers\"\u003eAI workstations and AI servers collection\u003c\/a\u003e. For configuration, research-grant or education enquiries, GST invoicing and lead times, email \u003ca href=\"mailto:sales@sourceit.com.sg\"\u003esales@sourceit.com.sg\u003c\/a\u003e quoting E6ZS0PT or E6ZS5PT.\u003c\/p\u003e\n\u003c\/div\u003e\u003c\/details\u003e\n\u003cdetails class=\"pd-acc\"\u003e\u003csummary style=\"cursor: pointer; font-weight: 700;\"\u003eDatasheet, Downloads \u0026amp; Setup Guide\u003c\/summary\u003e\u003cdiv class=\"pd-acc-body\"\u003e\n\u003cp\u003eOfficial HP and NVIDIA documentation:\u003c\/p\u003e\n\u003cul\u003e\n\u003cli\u003e\u003ca href=\"https:\/\/www.hp.com\/us-en\/workstations\/ai-stations\/zgx-fury-ai-station.html\" target=\"_blank\" rel=\"noopener\"\u003eHP ZGX Fury AI Station — official product overview\u003c\/a\u003e\u003c\/li\u003e\n\u003cli\u003e\u003ca href=\"https:\/\/www.hp.com\/sg-en\/products\/workstations\/product-details\/2104276823\" target=\"_blank\" rel=\"noopener\"\u003eHP ZGX Fury G1n E6ZS0PT — HP Singapore product page and specifications\u003c\/a\u003e\u003c\/li\u003e\n\u003cli\u003e\u003ca href=\"https:\/\/www.hp.com\/sg-en\/products\/workstations\/product-details\/2104276831\" target=\"_blank\" rel=\"noopener\"\u003eHP ZGX Fury G1n E6ZS5PT — HP Singapore product page and specifications\u003c\/a\u003e\u003c\/li\u003e\n\u003cli\u003e\u003ca href=\"https:\/\/h20195.www2.hp.com\/v2\/GetDocument.aspx?docname=c09316164\" target=\"_blank\" rel=\"noopener\"\u003eHP ZGX Fury AI Station G1n QuickSpecs (PDF)\u003c\/a\u003e\u003c\/li\u003e\n\u003cli\u003e\u003ca href=\"https:\/\/h20195.www2.hp.com\/v2\/GetPDF.aspx\/c09307810\" target=\"_blank\" rel=\"noopener\"\u003eHP ZGX Fury AI Station G1n datasheet (PDF)\u003c\/a\u003e\u003c\/li\u003e\n\u003cli\u003e\u003ca href=\"https:\/\/www.nvidia.com\/en-sg\/products\/workstations\/dgx-station\/\" target=\"_blank\" rel=\"noopener\"\u003eNVIDIA DGX Station (GB300) — official platform page (Singapore)\u003c\/a\u003e\u003c\/li\u003e\n\u003cli\u003e\u003ca href=\"https:\/\/www.nvidia.com\/content\/dam\/en-zz\/Solutions\/products\/workstations\/dgx-station\/workstation-datasheet-dgx-station-nvidia-us-5543600-web.pdf\" target=\"_blank\" rel=\"noopener\"\u003eNVIDIA DGX Station datasheet (PDF)\u003c\/a\u003e\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003c\/div\u003e\u003c\/details\u003e\n\u003cdetails class=\"pd-acc\"\u003e\u003csummary style=\"cursor: pointer; font-weight: 700;\"\u003eFrequently Asked Questions — HP ZGX Fury AI Station G1n\u003c\/summary\u003e\u003cdiv class=\"pd-acc-body\"\u003e\n\u003ch3\u003eWhat is the HP ZGX Fury AI Station G1n?\u003c\/h3\u003e\n\u003cp\u003eHP's build of the NVIDIA DGX Station GB300 platform: a liquid-cooled rackable tower with the GB300 Grace Blackwell Ultra Desktop Superchip, 748 GB coherent memory and up to 20 PFLOPS FP4 for training, fine-tuning and serving AI models of up to 1 trillion parameters on premises, backed by HP's 3-year onsite warranty.\u003c\/p\u003e\n\u003ch3\u003eWhat is the difference between the ZGX Fury G1n configurations?\u003c\/h3\u003e\n\u003cp\u003eOnly the display graphics card and the storage. E6ZS0PT#AB4 pairs an NVIDIA RTX PRO 2000 Blackwell with 4 × 2 TB NVMe; E6ZS5PT#AB4 steps up to an RTX PRO 4000 Blackwell SFF with 2 × 2 TB + 2 × 4 TB NVMe; the headless configuration has no discrete graphics card and 4 × 2 TB self-encrypting drives. The GB300 Superchip, 748 GB coherent memory, 72-core Grace CPU, ConnectX-8 networking and 3\/3\/3 warranty are identical across all three.\u003c\/p\u003e\n\u003ch3\u003eWhy would I choose the headless SED configuration?\u003c\/h3\u003e\n\u003cp\u003eIf the system is accessed remotely or sits in a shared compute room, a display GPU adds cost without value — and the self-encrypting drives give hardware-level encryption at rest, often required for regulated, government and financial-sector workloads in Singapore.\u003c\/p\u003e\n\u003ch3\u003eHow does the ZGX Fury G1n compare to the NVIDIA DGX Spark and HP ZGX Nano?\u003c\/h3\u003e\n\u003cp\u003eDifferent class entirely: the ZGX Fury delivers up to 20 PFLOPS FP4 and 748 GB coherent memory for models up to 1 trillion parameters, versus the GB10-based DGX Spark and ZGX Nano at 1 PFLOP and 128 GB (about 200 billion parameters). Choose the Nano or Spark for individual developers, the ZGX Fury for teams and production AI.\u003c\/p\u003e\n\u003ch3\u003eHP ZGX Fury G1n vs MSI XpertStation WS300, ASUS ET900N G3, GIGABYTE W775 or Supermicro — which should I choose?\u003c\/h3\u003e\n\u003cp\u003eAll are NVIDIA DGX Station GB300 systems with identical core compute. The ZGX Fury adds HP's published 3-year onsite warranty, a rackable liquid-cooled chassis with rails included, factory display-GPU options, up to 12 TB of HP Z Turbo Drive NVMe and the HP Z Runtime \/ Z Toolkit software. Talk to us about which fits your deployment and support expectations.\u003c\/p\u003e\n\u003ch3\u003eWhat AI models can it run?\u003c\/h3\u003e\n\u003cp\u003eHP cites development and fine-tuning of 100-billion-class models and inference on trillion-parameter-class models from the 748 GB coherent memory, which accommodates open-weight families such as Llama, Qwen, DeepSeek, Gemma, Mistral and gpt-oss at full scale.\u003c\/p\u003e\n\u003ch3\u003eCan multiple team members share one system?\u003c\/h3\u003e\n\u003cp\u003eYes — Multi-Instance GPU (MIG) partitions the Blackwell Ultra GPU into up to seven isolated instances for concurrent users and workloads.\u003c\/p\u003e\n\u003ch3\u003eCan two ZGX Fury systems be linked?\u003c\/h3\u003e\n\u003cp\u003eYes. The ConnectX-8 SuperNIC's two 400 Gb\/s QSFP112 ports link two DGX Station-class systems for more than 1.4 TB of coherent memory and up to 40 PFLOPS, per NVIDIA. SourceIT quotes the cabling with the systems.\u003c\/p\u003e\n\u003ch3\u003eDoes it run Windows?\u003c\/h3\u003e\n\u003cp\u003eNot today. The ZGX Fury ships with Ubuntu 24.04 LTS on an arm64 Grace CPU; HP states Windows support is coming soon and NVIDIA has announced DGX Station for Windows. Proprietary x86-only applications will not run natively.\u003c\/p\u003e\n\u003ch3\u003eDoes it need a server room?\u003c\/h3\u003e\n\u003cp\u003eNo rack is required — the ZGX Fury is liquid cooled inside its own tower chassis for 24\/7 operation, and it ships with rails for a 5U rack position later. Plan a dedicated 200–240 V AC circuit with a C20 inlet, as HP specifies.\u003c\/p\u003e\n\u003ch3\u003eWhat warranty does HP provide in Singapore?\u003c\/h3\u003e\n\u003cp\u003eHP's 3-year (3\/3\/3) limited warranty — 3 years of parts, labour and onsite repair — as stated on HP Singapore's product pages for E6ZS0PT and E6ZS5PT.\u003c\/p\u003e\n\u003ch3\u003eHow do I purchase the HP ZGX Fury G1n in Singapore?\u003c\/h3\u003e\n\u003cp\u003eSourceIT is an authorised IT reseller in Singapore and supplies the ZGX Fury G1n with GST invoice, HP onsite warranty and enterprise procurement assistance — contact \u003ca href=\"mailto:sales@sourceit.com.sg\"\u003esales@sourceit.com.sg\u003c\/a\u003e or use the Request for Quotation button for configuration options, lead times and pricing.\u003c\/p\u003e\n\u003c\/div\u003e\u003c\/details\u003e\n\u003ch3\u003eProduct Video\u003c\/h3\u003e\n\u003cp\u003e\u003ciframe width=\"560\" height=\"315\" src=\"https:\/\/www.youtube.com\/embed\/9hwxLrkR1p0\" title=\"HP ZGX Fury Launch | Z by HP\" allowfullscreen\u003e\u003c\/iframe\u003e\u003c\/p\u003e\n\u003c\/div\u003e\n","brand":"HP","offers":[{"title":"RTX PRO 2000 Blackwell \/ 4x2TB","offer_id":51421432807588,"sku":"E6ZS0PT","price":168838.0,"currency_code":"SGD","in_stock":true},{"title":"RTX PRO 4000 Blackwell \/ 2x2TB + 2x4TB","offer_id":51461812551844,"sku":"E6ZS5PT","price":172647.0,"currency_code":"SGD","in_stock":true},{"title":"No Graphics Card \/ 4x2TB SED","offer_id":51461812584612,"sku":"GPUonly","price":158683.0,"currency_code":"SGD","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0605\/0136\/0804\/files\/hp-zgx-fury-g1n-hero-square-v2.png?v=1790257801"},{"product_id":"gigabyte-w775-v10-l01-deskside-ai-supercomputer-with-nvidia-gb300-blackwell","title":"GIGABYTE W775-V10-L01 NVIDIA DGX Station GB300 AI Supercomputer","description":"\u003cdiv\u003e\n\u003cstyle\u003e\n.pd-acc { border: 1px solid #e2e2e2; border-radius: 8px; margin: 10px 0; overflow: hidden; }\n.pd-acc summary { cursor: pointer; padding: 14px 16px; font-weight: 700; font-size: 1.05em; list-style: none; display: flex; justify-content: space-between; align-items: center; background: #f7f7f7; }\n.pd-acc[open] \u003e summary { border-bottom: 1px solid #e2e2e2; }\n.pd-acc summary::-webkit-details-marker { display: none; }\n.pd-acc summary::after { content: \"+\"; font-size: 1.4em; font-weight: 400; line-height: 1; margin-left: 12px; }\n.pd-acc[open] \u003e summary::after { content: \"\\2212\"; }\n.pd-acc .pd-acc-body { padding: 8px 16px 14px; }\n.pd-spec, .pd-spec tr, .pd-spec td { border: none !important; background: transparent !important; }\n.pd-spec { width: 100%; border-collapse: collapse !important; margin: 6px 0 18px; }\n.pd-spec td { padding: 7px 10px 7px 0 !important; border-bottom: 1px solid #ececec !important; vertical-align: top; font-size: 0.95em; text-align: left; }\n.pd-spec td:first-child { width: 38%; font-weight: 600; color: #333; }\n.pd-spec tr:last-child td { border-bottom: none !important; }\n\u003c\/style\u003e\n\u003ch2\u003eGIGABYTE W775-V10-L01 NVIDIA DGX Station GB300 Grace Blackwell Ultra Deskside AI Supercomputer, 748GB Coherent Memory, 20 PFLOPS, Dual 400G ConnectX-8, Closed-Loop Liquid Cooling (W775-V10-L01) - Local Warranty\u003c\/h2\u003e\n\u003ch3\u003eGIGABYTE W775-V10-L01 — NVIDIA DGX Station GB300 Deskside AI Supercomputer\u003c\/h3\u003e\n\u003cp\u003eThe GIGABYTE W775-V10-L01 is GIGABYTE's deskside AI supercomputer built on NVIDIA DGX Station architecture around the NVIDIA GB300 Grace Blackwell Ultra Desktop Superchip. A 72-core NVIDIA Grace CPU (Arm Neoverse V2) and an NVIDIA Blackwell Ultra GPU share 748 GB of coherent memory over NVLink-C2C, delivering up to 20 petaFLOPS of FP4 AI compute and support for models of up to 1 trillion parameters — GIGABYTE's enterprise server engineering in a closed-loop liquid-cooled tower.\u003c\/p\u003e\n\u003cp\u003eFor Singapore enterprises, universities, government agencies and AI research labs, the W775-V10-L01 makes on-premise training, fine-tuning and inference of frontier-scale LLMs, agentic AI, computer vision and scientific simulation practical without a data-centre rack — with complete data sovereignty and no recurring cloud GPU cost.\u003c\/p\u003e\n\u003ch3\u003e748 GB Coherent Memory with ECC-Protected LPDDR5X\u003c\/h3\u003e\n\u003cp\u003e496 GB of LPDDR5X CPU memory with ECC at 396 GB\/s and 252 GB of HBM3E GPU memory at 7.1 TB\/s are unified over the 900 GB\/s NVLink-C2C link into a single 748 GB coherent memory space, so trillion-parameter models, long-context RAG pipelines and large simulation datasets stay resident without CPU–GPU offloading. Multi-Instance GPU (MIG) partitions the Blackwell Ultra GPU into up to seven isolated instances for shared team use.\u003c\/p\u003e\n\u003ch3\u003ePCIe Gen5 and Gen6 NVMe Storage, Three PCIe Gen5 Expansion Slots\u003c\/h3\u003e\n\u003cp\u003eTwo M.2 2280 PCIe Gen5 x4 slots off the Grace CPU support an OS RAID 1 pair, and two M.2 2280 PCIe Gen6 x4 slots off the ConnectX-8 provide ultra-fast dataset, checkpoint and vector-database storage; a further M.2 2230 E-key slot takes an optional Wi-Fi module. Three PCIe x16 slots — one with Gen5 x16 lanes for a display GPU and two with Gen5 x8 lanes — accept an NVIDIA RTX PRO Blackwell graphics card, additional NICs or accelerators.\u003c\/p\u003e\n\u003ch3\u003eDual 400 Gb\/s ConnectX-8 SuperNIC Networking and Onboard BMC Management\u003c\/h3\u003e\n\u003cp\u003eTwo 400 Gb\/s QSFP ports on the NVIDIA ConnectX-8 SuperNIC (up to 800 Gb\/s aggregate) support RDMA, GPUDirect RDMA and GPUDirect Storage for distributed training and private AI clusters, alongside a Marvell AQC113 10 Gb\/s LAN port and a dedicated 10\/100\/1000 Mb\/s management LAN. Video is integrated in the onboard BMC with a Mini-DisplayPort output (1920 × 1080 @ 60 Hz), a micro-USB UART port serves console access, and front and rear USB 3.2 Gen2 \/ Gen1 ports plus Realtek ALC4080 audio round out the I\/O.\u003c\/p\u003e\n\u003ch3\u003eClosed-Loop Liquid Cooling, 1600 W 80 PLUS Platinum, 245 × 500 × 531 mm Tower\u003c\/h3\u003e\n\u003cp\u003eA closed-loop liquid cooling solution with three 120 mm system fans keeps the Grace Blackwell Ultra Superchip at full performance through sustained 24 × 7 training and inference. A single 1600 W ATX 80 PLUS Platinum power supply (C19 power cord, 20 A circuit required) feeds the system. The tower measures 245 mm wide, 500.4 mm high and 531 mm deep, weighs 29.2 kg net (34.4 kg gross), and is rated for 10–35 °C operation at 8–80 % non-condensing humidity. It ships with Ubuntu with NVIDIA AI Developer Tools and carries Powered by NVIDIA and Ubuntu 24.04 LTS Certified Hardware badges.\u003c\/p\u003e\n\u003ch3\u003eAvailable in Singapore Through SourceIT\u003c\/h3\u003e\n\u003cp\u003eSourceIT supplies the GIGABYTE W775-V10-L01 with a full GST invoice, local warranty support and enterprise procurement assistance. GIGABYTE states that warranty terms vary by product and region; SourceIT confirms the applicable GIGABYTE enterprise warranty term for Singapore in writing at quotation. For desktop-scale AI, see the NVIDIA DGX Spark or GIGABYTE AI TOP ATOM; for alternative GB300 stations, see the HP ZGX Fury G1n, MSI XpertStation WS300 and ASUS ExpertCenter Pro ET900N G3.\u003c\/p\u003e\n\u003cdetails class=\"pd-acc\"\u003e\u003csummary style=\"cursor: pointer; font-weight: 700;\"\u003eGIGABYTE W775-V10-L01 Datasheet — Technical Specifications\u003c\/summary\u003e\u003cdiv class=\"pd-acc-body\"\u003e\n\u003ch3\u003eGeneral\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eProduct Name\u003c\/td\u003e\n\u003ctd\u003eGIGABYTE W775-V10-L01 Deskside AI Supercomputer\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eModel \/ Part Number\u003c\/td\u003e\n\u003ctd\u003eW775-V10-L01 (GIGABYTE P\/N 6NW775V10MR000L01)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eCategory\u003c\/td\u003e\n\u003ctd\u003eDeskside AI supercomputer \/ enterprise AI workstation (NVIDIA DGX Station GB300 platform)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eForm Factor\u003c\/td\u003e\n\u003ctd\u003eTower\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eManufacturer Support\u003c\/td\u003e\n\u003ctd\u003eGIGABYTE enterprise warranty — term varies by region and is confirmed at quotation\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eLocal Warranty (Singapore)\u003c\/td\u003e\n\u003ctd\u003eLocal warranty support, procurement and GST invoice by SourceIT\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\u003ch3\u003eProcessor \u0026amp; GPU\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eCompute Platform\u003c\/td\u003e\n\u003ctd\u003eNVIDIA DGX Station architecture — NVIDIA GB300 Grace Blackwell Ultra Desktop Superchip\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eCPU\u003c\/td\u003e\n\u003ctd\u003e1 × NVIDIA Grace CPU, 72 Arm Neoverse V2 cores\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eGPU\u003c\/td\u003e\n\u003ctd\u003e1 × NVIDIA Blackwell Ultra GPU (fifth-generation Tensor Cores, second-generation Transformer Engine)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eCPU–GPU Interconnect\u003c\/td\u003e\n\u003ctd\u003eNVIDIA NVLink-C2C, 900 GB\/s (per NVIDIA DGX Station datasheet)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eAI Performance\u003c\/td\u003e\n\u003ctd\u003eUp to 20 PFLOPS FP4 Tensor Core (with sparsity), per NVIDIA — 10 PFLOPS FP8\/FP6, 5 PFLOPS FP16\/BF16\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eModel Support\u003c\/td\u003e\n\u003ctd\u003eAI models up to 1 trillion parameters, per NVIDIA\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eGPU Partitioning\u003c\/td\u003e\n\u003ctd\u003eMulti-Instance GPU (MIG) — up to 7 instances\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\u003ch3\u003eMemory\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eTotal Coherent Memory\u003c\/td\u003e\n\u003ctd\u003eUp to 748 GB unified CPU + GPU memory space\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eGPU Memory\u003c\/td\u003e\n\u003ctd\u003e252 GB HBM3E, 7.1 TB\/s\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eCPU Memory\u003c\/td\u003e\n\u003ctd\u003e496 GB LPDDR5X with ECC, 396 GB\/s\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\u003ch3\u003eStorage\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eOS Storage\u003c\/td\u003e\n\u003ctd\u003e2 × M.2 2280 PCIe Gen5 x4 (from CPU), OS RAID 1 support\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eData Storage\u003c\/td\u003e\n\u003ctd\u003e2 × M.2 2280 PCIe Gen6 x4 (from ConnectX-8)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eWireless Slot\u003c\/td\u003e\n\u003ctd\u003e1 × M.2 2230 E-key PCIe Gen2 x1 (Wi-Fi module optional)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eInstalled Drives\u003c\/td\u003e\n\u003ctd\u003eSourceIT-supplied configuration: 4 × 1.92 TB NVMe M.2 SSD — confirmed at quotation (GIGABYTE publishes no factory drive configuration)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\u003ch3\u003eExpansion\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003ePCIe Slots\u003c\/td\u003e\n\u003ctd\u003e1 × PCIe x16 (Gen5 x16 lanes, for GPU); 2 × PCIe x16 (Gen5 x8 lanes)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\u003ch3\u003eNetworking\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eHigh-Speed Networking\u003c\/td\u003e\n\u003ctd\u003e2 × 400 Gb\/s QSFP ports via NVIDIA ConnectX-8 SuperNIC (up to 800 Gb\/s)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eGeneral Ethernet\u003c\/td\u003e\n\u003ctd\u003e1 × 10 Gb\/s LAN (Marvell AQC113)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eManagement LAN\u003c\/td\u003e\n\u003ctd\u003e1 × 10\/100\/1000 Mb\/s MLAN\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eWireless\u003c\/td\u003e\n\u003ctd\u003e2 × Wi-Fi antenna ports (module not stated as included)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\u003ch3\u003eI\/O Ports\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eFront I\/O\u003c\/td\u003e\n\u003ctd\u003e1 × USB 3.2 Gen2 Type-C, 2 × USB 3.2 Gen1 Type-A, 2 × audio jacks, power button with LED, reset button\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eRear I\/O\u003c\/td\u003e\n\u003ctd\u003e4 × USB 3.2 Gen2 Type-A, 1 × micro USB (USB-to-UART), 1 × Mini-DP, 2 × QSFP, 1 × RJ45 10G, 1 × MLAN, 3 × audio jacks\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eVideo\u003c\/td\u003e\n\u003ctd\u003eIntegrated in onboard BMC — 1 × Mini-DisplayPort, 1920 × 1080 @ 60 Hz\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eAudio\u003c\/td\u003e\n\u003ctd\u003eRealtek ALC4080 codec\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\u003ch3\u003eOperating System \u0026amp; Software\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eOperating System\u003c\/td\u003e\n\u003ctd\u003eUbuntu with NVIDIA AI Developer Tools; Ubuntu 24.04 LTS Certified Hardware\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003ePlatform Badges\u003c\/td\u003e\n\u003ctd\u003ePowered by NVIDIA; Ubuntu Certified\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\u003ch3\u003ePower, Cooling, Physical \u0026amp; Environmental\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003ePower Supply\u003c\/td\u003e\n\u003ctd\u003eSingle 1600 W ATX 80 PLUS Platinum; C19 power cord; 20 A circuit required\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eCooling\u003c\/td\u003e\n\u003ctd\u003eClosed-loop liquid cooling; 3 × 120 × 120 × 25 mm system fans\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eDimensions (W × H × D)\u003c\/td\u003e\n\u003ctd\u003e245 × 500.4 × 531 mm; packaging 732 × 400 × 775 mm\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eWeight\u003c\/td\u003e\n\u003ctd\u003e29.2 kg net; 34.4 kg gross\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eOperating Environment\u003c\/td\u003e\n\u003ctd\u003e10 – 35 °C, 8 – 80 % non-condensing; non-operating −40 – 60 °C, 20 – 95 %\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\u003c\/div\u003e\u003c\/details\u003e\n\u003cdetails class=\"pd-acc\"\u003e\u003csummary style=\"cursor: pointer; font-weight: 700;\"\u003eCompatibility \u0026amp; Software Support\u003c\/summary\u003e\u003cdiv class=\"pd-acc-body\"\u003e\n\u003ch3\u003eNVIDIA AI Software Stack\u003c\/h3\u003e\n\u003cp\u003eShips with Ubuntu Linux and the NVIDIA AI Developer Tools preinstalled — the same NVIDIA software foundation used across the DGX family. CUDA, cuDNN, TensorRT and TensorRT-LLM, NVIDIA NeMo, NVIDIA NIM microservices, Triton Inference Server, RAPIDS and NGC containers run natively on the Grace Blackwell Ultra platform, alongside PyTorch, TensorFlow, JAX, vLLM, Hugging Face Transformers, DeepSpeed and ONNX Runtime. Docker and Kubernetes deployment are supported for containerised MLOps pipelines. NVIDIA AI Enterprise is available as a separate subscription for supported production deployments.\u003c\/p\u003e\n\u003ch3\u003eArm64 Architecture — Plan Your Software Accordingly\u003c\/h3\u003e\n\u003cp\u003eThe Grace CPU is an Arm Neoverse V2 processor, so the operating system, drivers and container images are arm64 (aarch64) builds. Mainstream AI frameworks and NVIDIA NGC containers are published for arm64, but proprietary x86-only applications will not run natively — check any commercial software you rely on before ordering. NVIDIA has announced a separate DGX Station for Windows SKU (Windows plus Windows Subsystem for Linux) for a later release; the units SourceIT supplies today ship with Ubuntu.\u003c\/p\u003e\n\u003ch3\u003eClustering and Networking Compatibility\u003c\/h3\u003e\n\u003cp\u003eThe two NVIDIA ConnectX-8 SuperNIC QSFP112 ports run at 400 Gb\/s each (up to 800 Gb\/s aggregate) with RDMA over Converged Ethernet, GPUDirect RDMA and GPUDirect Storage support, so two or more systems can be linked for distributed training, or connected to an NVMe-over-fabric storage array. QSFP112 transceivers or direct-attach cables are ordered separately to match your switch fabric — SourceIT will quote compatible optics alongside the system. The 10GbE RJ45 port covers everyday site networking, and the dedicated management LAN keeps out-of-band BMC traffic isolated from production data.\u003c\/p\u003e\n\u003ch3\u003eGIGABYTE Enterprise Platform\u003c\/h3\u003e\n\u003cp\u003eThe W775-V10-L01 is engineered by GIGABYTE's enterprise (Giga Computing) division and sits alongside GIGABYTE's NVIDIA-based rack servers, so it integrates with standard BMC-based management workflows. GIGABYTE reports peak throughput of up to 9,000 tokens per second and a typical 1,000–1,200 W power draw under load on this platform.\u003c\/p\u003e\n\u003c\/div\u003e\u003c\/details\u003e\n\u003cdetails class=\"pd-acc\"\u003e\u003csummary style=\"cursor: pointer; font-weight: 700;\"\u003ePackage Contents \u0026amp; Ordering Information\u003c\/summary\u003e\u003cdiv class=\"pd-acc-body\"\u003e\n\u003ch3\u003eWhat Ships With the System\u003c\/h3\u003e\n\u003cul\u003e\n\u003cli\u003eGIGABYTE W775-V10-L01 deskside AI supercomputer with NVIDIA GB300 Grace Blackwell Ultra Desktop Superchip\u003c\/li\u003e\n\u003cli\u003e1600 W 80 PLUS Platinum power supply (installed) and C19 power cord\u003c\/li\u003e\n\u003cli\u003eNVMe M.2 storage as per the SourceIT-supplied configuration (4 × 1.92 TB, confirmed at quotation)\u003c\/li\u003e\n\u003cli\u003eQuick installation guide and GIGABYTE warranty \/ enterprise support documentation\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003cp\u003eQSFP112 transceivers or DAC cables, a Wi-Fi module, additional NVMe SSDs and an optional NVIDIA RTX PRO Blackwell display GPU are quoted separately.\u003c\/p\u003e\n\u003ch3\u003eNVIDIA DGX Station GB300 Systems at SourceIT Singapore\u003c\/h3\u003e\n\u003cp\u003eThe GIGABYTE W775-V10-L01 is quoted by SourceIT Singapore with GST invoice; drive configuration and GIGABYTE enterprise warranty terms are confirmed in writing at quotation.\u003c\/p\u003e\n\u003cul\u003e\n\u003cli\u003e\n\u003ca href=\"\/products\/asus-expertcenter-pro-et900n-g3-gb300-ai-supercomputer\"\u003eASUS ExpertCenter Pro ET900N G3\u003c\/a\u003e — NVIDIA DGX Station GB300 platform, same 748 GB \/ 20 PFLOPS core compute\u003c\/li\u003e\n\u003cli\u003e\n\u003ca href=\"\/products\/msi-xpertstation-ws300-nvidia-dgx-station-ai-supercomputer\"\u003eMSI XpertStation WS300\u003c\/a\u003e — NVIDIA DGX Station GB300 platform, same 748 GB \/ 20 PFLOPS core compute\u003c\/li\u003e\n\u003cli\u003e\n\u003ca href=\"\/products\/hp-zgx-fury-g1n-gb300-ai-workstation-with-nvidia-blackwell-ultra\"\u003eHP ZGX Fury AI Station G1n\u003c\/a\u003e — NVIDIA DGX Station GB300 platform, same 748 GB \/ 20 PFLOPS core compute\u003c\/li\u003e\n\u003cli\u003e\n\u003ca href=\"\/products\/nvidia-dgx-spark-ai-supercomputer-4tb-940-54242-0007-000\"\u003eNVIDIA DGX Spark\u003c\/a\u003e — desktop-scale GB10 system (128 GB unified memory, 1 PFLOP FP4) for individual developers\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003cp\u003eBrowse the full \u003ca href=\"\/collections\/nvidia-dgx-spark\"\u003eNVIDIA DGX Spark \u0026amp; DGX Station collection\u003c\/a\u003e. For volume, project, research-grant or education pricing, GST invoicing and current lead times, email \u003ca href=\"mailto:sales@sourceit.com.sg\"\u003esales@sourceit.com.sg\u003c\/a\u003e.\u003c\/p\u003e\n\u003c\/div\u003e\u003c\/details\u003e\n\u003cdetails class=\"pd-acc\"\u003e\u003csummary style=\"cursor: pointer; font-weight: 700;\"\u003eDatasheet \u0026amp; Downloads\u003c\/summary\u003e\u003cdiv class=\"pd-acc-body\"\u003e\n\u003cp\u003eOfficial GIGABYTE documentation for the W775-V10-L01:\u003c\/p\u003e\n\u003cul\u003e\n\u003cli\u003e\u003ca href=\"https:\/\/download.gigabyte.com\/FileList\/DataSheet\/GIGABYTE%20W775-V10-L01_datasheet.pdf\" target=\"_blank\" rel=\"noopener\"\u003eGIGABYTE W775-V10-L01 datasheet (PDF)\u003c\/a\u003e\u003c\/li\u003e\n\u003cli\u003e\u003ca href=\"https:\/\/www.gigabyte.com\/sg\/Enterprise\/Tower-Server\/W775-V10-L01\" target=\"_blank\" rel=\"noopener\"\u003eGIGABYTE W775-V10-L01 — product page (Singapore)\u003c\/a\u003e\u003c\/li\u003e\n\u003cli\u003e\u003ca href=\"https:\/\/www.gigabyte.com\/Enterprise\/Tower-Server\/W775-V10-L01#Specifications\" target=\"_blank\" rel=\"noopener\"\u003eGIGABYTE W775-V10-L01 — full specifications\u003c\/a\u003e\u003c\/li\u003e\n\u003cli\u003e\u003ca href=\"https:\/\/www.nvidia.com\/en-sg\/products\/workstations\/dgx-station\/\" target=\"_blank\" rel=\"noopener\"\u003eNVIDIA DGX Station (GB300) — official platform page (Singapore)\u003c\/a\u003e\u003c\/li\u003e\n\u003cli\u003e\u003ca href=\"https:\/\/www.nvidia.com\/content\/dam\/en-zz\/Solutions\/products\/workstations\/dgx-station\/workstation-datasheet-dgx-station-nvidia-us-5543600-web.pdf\" target=\"_blank\" rel=\"noopener\"\u003eNVIDIA DGX Station datasheet (PDF)\u003c\/a\u003e\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003c\/div\u003e\u003c\/details\u003e\n\u003cdetails class=\"pd-acc\"\u003e\u003csummary style=\"cursor: pointer; font-weight: 700;\"\u003eFrequently Asked Questions — GIGABYTE W775-V10-L01\u003c\/summary\u003e\u003cdiv class=\"pd-acc-body\"\u003e\n\u003ch3\u003eWhat is the GIGABYTE W775-V10-L01?\u003c\/h3\u003e\n\u003cp\u003eGIGABYTE's deskside AI supercomputer on the NVIDIA DGX Station GB300 platform: a 72-core Grace CPU and Blackwell Ultra GPU sharing 748 GB of coherent memory, up to 20 PFLOPS FP4, closed-loop liquid cooled, in a 29 kg tower — for training, fine-tuning and serving AI models of up to 1 trillion parameters on premises.\u003c\/p\u003e\n\u003ch3\u003eIs this a GIGABYTE AI TOP product?\u003c\/h3\u003e\n\u003cp\u003eNo. GIGABYTE lists the W775-V10-L01 under its enterprise Tower Server range, engineered by Giga Computing; AI TOP is GIGABYTE's brand for consumer and prosumer AI desktops such as the GB10-based AI TOP ATOM, which SourceIT also stocks.\u003c\/p\u003e\n\u003ch3\u003eHow does it compare to the NVIDIA DGX Spark or GIGABYTE AI TOP ATOM?\u003c\/h3\u003e\n\u003cp\u003eDifferent class entirely: the W775-V10-L01 delivers up to 20 PFLOPS FP4 and 748 GB coherent memory for models up to 1 trillion parameters, versus the GB10-based DGX Spark and AI TOP ATOM at 1 PFLOP and 128 GB (about 200 billion parameters). Choose a GB10 system for individual developers; the W775-V10-L01 for teams and production-scale AI.\u003c\/p\u003e\n\u003ch3\u003eWhat storage does it come with?\u003c\/h3\u003e\n\u003cp\u003eGIGABYTE supplies the platform with four M.2 2280 slots (two PCIe Gen5 x4 with OS RAID 1 support, two PCIe Gen6 x4) and publishes no factory drive configuration. Units supplied by SourceIT are configured with 4 × 1.92 TB NVMe M.2 SSDs; confirm the drive layout at quotation.\u003c\/p\u003e\n\u003ch3\u003eIs the W775-V10-L01 liquid cooled?\u003c\/h3\u003e\n\u003cp\u003eYes — a closed-loop liquid cooling solution with three 120 mm system fans is built into the tower, so no external liquid-cooled rack is required.\u003c\/p\u003e\n\u003ch3\u003eCan multiple team members share one system?\u003c\/h3\u003e\n\u003cp\u003eYes — Multi-Instance GPU (MIG) partitions the Blackwell Ultra GPU into up to seven isolated instances for concurrent users and workloads.\u003c\/p\u003e\n\u003ch3\u003eWhat power and environment does it need?\u003c\/h3\u003e\n\u003cp\u003eA single 1600 W 80 PLUS Platinum PSU on a dedicated 20 A circuit with a C19 power cord, in an air-conditioned space within 10–35 °C. GIGABYTE reports typical draw of 1,000–1,200 W under load.\u003c\/p\u003e\n\u003ch3\u003eWhat warranty does GIGABYTE provide?\u003c\/h3\u003e\n\u003cp\u003eGIGABYTE states that warranty terms vary by product and region and does not publish a fixed term for the W775-V10-L01. SourceIT confirms the applicable GIGABYTE enterprise warranty for Singapore in writing at quotation and provides local warranty support.\u003c\/p\u003e\n\u003ch3\u003eHow do I purchase the W775-V10-L01 in Singapore?\u003c\/h3\u003e\n\u003cp\u003eSourceIT is an authorised IT reseller in Singapore and supplies the GIGABYTE W775-V10-L01 with GST invoice, local warranty support and enterprise procurement assistance — contact sales@sourceit.com.sg for configuration options, volume pricing and lead times.\u003c\/p\u003e\n\u003c\/div\u003e\u003c\/details\u003e\n\u003ch3\u003eProduct Video\u003c\/h3\u003e\n\u003cp\u003e\u003ciframe width=\"560\" height=\"315\" src=\"https:\/\/www.youtube.com\/embed\/mTWxEE-GxmU\" title=\"GIGABYTE Enterprise AI Platforms | Rack-Scale \u0026amp; AI Workstations in Action | Giga Computing\" frameborder=\"0\" allow=\"accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture\" allowfullscreen\u003e\u003c\/iframe\u003e\u003c\/p\u003e\n\u003c\/div\u003e","brand":"Nvidia","offers":[{"title":"Default Title","offer_id":51425918124196,"sku":"W775-V10-L01","price":165000.0,"currency_code":"SGD","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0605\/0136\/0804\/files\/gigabyte-w775-v10-l01-gb300-tower-square.png?v=1790257071"},{"product_id":"supermicro-super-ai-station-ars-511gd-nb-lcc-nvidia-dgx-station-gb300","title":"Supermicro Super AI Station ARS-511GD-NB-LCC NVIDIA DGX Station GB300 AI Workstation","description":"\u003cdiv\u003e\n\u003cstyle\u003e\n.pd-acc { border: 1px solid #e2e2e2; border-radius: 8px; margin: 10px 0; overflow: hidden; }\n.pd-acc summary { cursor: pointer; padding: 14px 16px; font-weight: 700; font-size: 1.05em; list-style: none; display: flex; justify-content: space-between; align-items: center; background: #f7f7f7; }\n.pd-acc[open] \u003e summary { border-bottom: 1px solid #e2e2e2; }\n.pd-acc summary::-webkit-details-marker { display: none; }\n.pd-acc summary::after { content: \"+\"; font-size: 1.4em; font-weight: 400; line-height: 1; margin-left: 12px; }\n.pd-acc[open] \u003e summary::after { content: \"\\2212\"; }\n.pd-acc .pd-acc-body { padding: 8px 16px 14px; }\n.pd-spec, .pd-spec tr, .pd-spec td { border: none !important; background: transparent !important; }\n.pd-spec { width: 100%; border-collapse: collapse !important; margin: 6px 0 18px; }\n.pd-spec td { padding: 7px 10px 7px 0 !important; border-bottom: 1px solid #ececec !important; vertical-align: top; font-size: 0.95em; text-align: left; }\n.pd-spec td:first-child { width: 38%; font-weight: 600; color: #333; }\n.pd-spec tr:last-child td { border-bottom: none !important; }\n\u003c\/style\u003e\n\u003ch2\u003eSupermicro Super AI Station ARS-511GD-NB-LCC NVIDIA DGX Station GB300 Grace Blackwell Ultra AI Workstation, 748GB Coherent Memory, 20 PFLOPS, 4 × 1.92TB NVMe, Dual 400G ConnectX-8, Closed-Loop Liquid Cooling (ARS-511GD-NB-LCC) - 3 Years Local Warranty\u003c\/h2\u003e\n\u003ch3\u003eSupermicro Super AI Station — NVIDIA DGX Station GB300 in a Tower or 5U Rack Chassis\u003c\/h3\u003e\n\u003cp\u003eThe Supermicro Super AI Station ARS-511GD-NB-LCC is Supermicro's build of the NVIDIA DGX Station GB300 platform: an NVIDIA GB300 Grace Blackwell Ultra Desktop Superchip pairing a 72-core NVIDIA Grace CPU (Arm Neoverse V2) with an NVIDIA Blackwell Ultra B300 GPU over NVLink-C2C, sharing 748 GB of coherent memory and delivering up to 20 petaFLOPS of FP4 AI performance for training, fine-tuning and serving AI models of up to 1 trillion parameters — on premises, under your own data governance.\u003c\/p\u003e\n\u003cp\u003eSupermicro brings its data-centre server engineering to the deskside: closed-loop direct-to-chip liquid cooling, a 1600 W Titanium power supply, IPMI 2.0 and Redfish remote management, a Silicon Root of Trust and a chassis that works as a tower today and converts to a 5U rack unit with an optional rail kit when your AI deployment grows. SourceIT supplies it in Singapore with Supermicro's 3-year parts-and-service warranty with next-business-day onsite hardware support, a full GST invoice and enterprise procurement assistance.\u003c\/p\u003e\n\u003ch3\u003e748 GB ECC Coherent Memory for Trillion-Parameter Models\u003c\/h3\u003e\n\u003cp\u003e496 GB of ECC LPDDR5X CPU memory (four 128 GB SOCAMM modules) and 252 GB of ECC HBM3e GPU memory unite over NVLink-C2C into a single 748 GB coherent memory space, so frontier-scale LLMs, long-context RAG pipelines, fine-tuning runs and simulation datasets stay resident without CPU–GPU offloading. Multi-Instance GPU (MIG) partitions the Blackwell Ultra GPU into up to seven isolated instances so several users or workloads share one system with guaranteed resources.\u003c\/p\u003e\n\u003ch3\u003e4 × 1.92 TB Enterprise NVMe, Ubuntu 24.04 LTS Preinstalled\u003c\/h3\u003e\n\u003cp\u003eThe SourceIT configuration ships with four 1.92 TB enterprise NVMe M.2 SSDs (1 DWPD): two in the PCIe 5.0 x4 slots as an OS RAID 1 pair and two in the PCIe 6.0 x4 slots as a RAID 0 data volume, both configured through the host OS. Ubuntu 24.04 LTS with NVIDIA drivers, CUDA and cuDNN is preinstalled alongside the NVIDIA AI Developer Tools, so the system is ready for vLLM, NVIDIA NeMo, NIM microservices and the wider CUDA ecosystem on delivery. An M.2 E-key slot takes an optional Wi-Fi \/ Bluetooth module.\u003c\/p\u003e\n\u003ch3\u003eDual 400 GbE ConnectX-8 Networking, PCIe 5.0 Expansion and an Optional Second GPU\u003c\/h3\u003e\n\u003cp\u003eTwo QSFP 400 GbE ports on the NVIDIA ConnectX-8 SuperNIC (up to 800 Gb\/s aggregate) connect the Super AI Station into private AI clusters, distributed training fabrics and NVMe-over-fabric storage, with a 10 GbE RJ45 port for site networking and a dedicated 1 GbE BMC port for out-of-band management. A PCIe 5.0 x16 full-height double-width slot accepts an optional NVIDIA RTX PRO Blackwell display GPU (RTX PRO 2000, 4000 SFF, 6000 Max-Q or 6000 Workstation Edition), and two PCIe 5.0 x8 slots take additional NICs or accelerators.\u003c\/p\u003e\n\u003ch3\u003eEnterprise Management, Security and Serviceability\u003c\/h3\u003e\n\u003cp\u003eAn onboard BMC provides IPMI 2.0 with KVM-over-LAN and virtual media plus a Redfish API, so the system slots into existing data-centre management tooling. TPM 2.0 and a Silicon Root of Trust compliant with NIST 800-193 protect firmware integrity, and chassis intrusion detection is built in. Closed-loop direct-to-chip liquid cooling with five system fans keeps the Superchip at full performance through sustained 24 × 7 training and inference.\u003c\/p\u003e\n\u003ch3\u003e1600 W Titanium Power, Tower or 5U, 40 kg\u003c\/h3\u003e\n\u003cp\u003eA single 1600 W 80 PLUS Titanium (94 %) power supply feeds the system on 115–240 V AC. The chassis measures 218.4 mm high, 454.7 mm wide and 701 mm deep, weighs 40 kg net (51 kg gross), and is rated for 15–32 °C operation at 20–80 % non-condensing humidity. The optional MCP-290-75905-0B rail kit converts it to a 5U rackmount unit.\u003c\/p\u003e\n\u003cdetails class=\"pd-acc\"\u003e\u003csummary style=\"cursor: pointer; font-weight: 700;\"\u003eSupermicro Super AI Station ARS-511GD-NB-LCC Datasheet — Technical Specifications\u003c\/summary\u003e\u003cdiv class=\"pd-acc-body\"\u003e\n\u003ch3\u003eGeneral\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eProduct Name\u003c\/td\u003e\n\u003ctd\u003eSupermicro Super AI Station\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eModel \/ Part Number\u003c\/td\u003e\n\u003ctd\u003eARS-511GD-NB-LCC (motherboard Super GPU-NVGB300-WSHPM; chassis CSE-759TS-R0CNBP)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eCategory\u003c\/td\u003e\n\u003ctd\u003eEnterprise AI workstation \/ deskside AI supercomputer (NVIDIA DGX Station GB300 platform)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eForm Factor\u003c\/td\u003e\n\u003ctd\u003eTower, convertible to 5U rackmount with optional rail kit MCP-290-75905-0B\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eWarranty\u003c\/td\u003e\n\u003ctd\u003e3 years parts and service with next-business-day onsite hardware support by Supermicro (OSNBD3)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eLocal Warranty (Singapore)\u003c\/td\u003e\n\u003ctd\u003e3 Years Supermicro NBD onsite; procurement and GST invoice by SourceIT\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\u003ch3\u003eProcessor \u0026amp; GPU\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eCompute Platform\u003c\/td\u003e\n\u003ctd\u003eNVIDIA DGX Station architecture — NVIDIA GB300 Grace Blackwell Ultra Desktop Superchip\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eCPU\u003c\/td\u003e\n\u003ctd\u003e1 × NVIDIA Grace CPU, 72 Arm Neoverse V2 cores\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eGPU\u003c\/td\u003e\n\u003ctd\u003e1 × NVIDIA Blackwell Ultra GPU (fifth-generation Tensor Cores, second-generation Transformer Engine)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eCPU–GPU Interconnect\u003c\/td\u003e\n\u003ctd\u003eNVIDIA NVLink-C2C, 900 GB\/s (per NVIDIA DGX Station datasheet)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eAI Performance\u003c\/td\u003e\n\u003ctd\u003eUp to 20 PFLOPS FP4 Tensor Core (with sparsity), per NVIDIA — 10 PFLOPS FP8\/FP6, 5 PFLOPS FP16\/BF16\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eModel Support\u003c\/td\u003e\n\u003ctd\u003eAI models up to 1 trillion parameters, per NVIDIA\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eGPU Partitioning\u003c\/td\u003e\n\u003ctd\u003eMulti-Instance GPU (MIG) — up to 7 instances\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eOptional Second GPU\u003c\/td\u003e\n\u003ctd\u003e1 × double-width PCIe 5.0 x16 slot for NVIDIA RTX PRO 2000 \/ 4000 SFF \/ 6000 Max-Q \/ 6000 Workstation Edition Blackwell (quoted separately)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\u003ch3\u003eMemory\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eTotal Coherent Memory\u003c\/td\u003e\n\u003ctd\u003e748 GB unified CPU + GPU memory space\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eGPU Memory\u003c\/td\u003e\n\u003ctd\u003e252 GB ECC HBM3e\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eCPU Memory\u003c\/td\u003e\n\u003ctd\u003e496 GB ECC LPDDR5X, 4 × 128 GB SOCAMM\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\u003ch3\u003eStorage\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eOS Storage\u003c\/td\u003e\n\u003ctd\u003e2 × M.2 2280 PCIe 5.0 x4 NVMe slots — fitted with 2 × 1.92 TB enterprise NVMe Gen4 M.2 SSD (1 DWPD), RAID 1 via OS\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eData Storage\u003c\/td\u003e\n\u003ctd\u003e2 × M.2 2280 PCIe 6.0 x4 NVMe slots — fitted with 2 × 1.92 TB enterprise NVMe Gen4 M.2 SSD (1 DWPD), RAID 0 via OS\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eWireless Slot\u003c\/td\u003e\n\u003ctd\u003e1 × M.2 2230 E-key PCIe 2.0 x1 (Wi-Fi \/ Bluetooth module optional)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\u003ch3\u003eExpansion\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003ePCIe Slots\u003c\/td\u003e\n\u003ctd\u003e1 × PCIe 5.0 x16 FHFL double-width; 2 × PCIe 5.0 x8 (in x16) FHHL\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\u003ch3\u003eNetworking\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eHigh-Speed Networking\u003c\/td\u003e\n\u003ctd\u003e2 × QSFP 400 GbE, NVIDIA ConnectX-8 SuperNIC (up to 800 Gb\/s)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eGeneral Ethernet\u003c\/td\u003e\n\u003ctd\u003e1 × RJ45 10 GbE\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eManagement LAN\u003c\/td\u003e\n\u003ctd\u003e1 × RJ45 1 GbE dedicated BMC port\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\u003ch3\u003eI\/O Ports\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eFront I\/O\u003c\/td\u003e\n\u003ctd\u003e2 × USB 3.0 Type-A\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eRear I\/O\u003c\/td\u003e\n\u003ctd\u003e4 × USB 3.0 Gen2 Type-A, 1 × mini-DisplayPort, audio\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\u003ch3\u003eManagement \u0026amp; Security\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eBMC\u003c\/td\u003e\n\u003ctd\u003eIPMI 2.0 with KVM-over-LAN and virtual media over LAN; Redfish API\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eSecurity\u003c\/td\u003e\n\u003ctd\u003eTPM 2.0; Silicon Root of Trust (NIST 800-193 compliant); chassis intrusion detection\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eBIOS\u003c\/td\u003e\n\u003ctd\u003eAMI, 64 MB SPI flash\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\u003ch3\u003eOperating System \u0026amp; Software\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eOperating System\u003c\/td\u003e\n\u003ctd\u003eUbuntu 24.04 LTS with NVIDIA AI Developer Tools — preinstalled with NVIDIA drivers, CUDA and cuDNN\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\u003ch3\u003ePower, Cooling, Physical \u0026amp; Environmental\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003ePower Supply\u003c\/td\u003e\n\u003ctd\u003e1 × 1600 W 80 PLUS Titanium (94 %); input 115–240 V AC, 50–60 Hz\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eCooling\u003c\/td\u003e\n\u003ctd\u003eClosed-loop direct-to-chip liquid cooling; fans 2 × 140 mm, 2 × 80 mm, 1 × 60 mm\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eDimensions (H × W × D)\u003c\/td\u003e\n\u003ctd\u003e218.4 × 454.7 × 701 mm (8.6 × 17.9 × 27.6 in)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eWeight\u003c\/td\u003e\n\u003ctd\u003e40 kg net; 51 kg gross\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eOperating Environment\u003c\/td\u003e\n\u003ctd\u003e15 – 32 °C, 20 – 80 % non-condensing\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eColour\u003c\/td\u003e\n\u003ctd\u003eBlack\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\u003c\/div\u003e\u003c\/details\u003e\n\u003cdetails class=\"pd-acc\"\u003e\u003csummary style=\"cursor: pointer; font-weight: 700;\"\u003eCompatibility \u0026amp; Software Support\u003c\/summary\u003e\u003cdiv class=\"pd-acc-body\"\u003e\n\u003ch3\u003eNVIDIA AI Software Stack\u003c\/h3\u003e\n\u003cp\u003eShips with Ubuntu Linux and the NVIDIA AI Developer Tools preinstalled — the same NVIDIA software foundation used across the DGX family. CUDA, cuDNN, TensorRT and TensorRT-LLM, NVIDIA NeMo, NVIDIA NIM microservices, Triton Inference Server, RAPIDS and NGC containers run natively on the Grace Blackwell Ultra platform, alongside PyTorch, TensorFlow, JAX, vLLM, Hugging Face Transformers, DeepSpeed and ONNX Runtime. Docker and Kubernetes deployment are supported for containerised MLOps pipelines. NVIDIA AI Enterprise is available as a separate subscription for supported production deployments.\u003c\/p\u003e\n\u003ch3\u003eArm64 Architecture — Plan Your Software Accordingly\u003c\/h3\u003e\n\u003cp\u003eThe Grace CPU is an Arm Neoverse V2 processor, so the operating system, drivers and container images are arm64 (aarch64) builds. Mainstream AI frameworks and NVIDIA NGC containers are published for arm64, but proprietary x86-only applications will not run natively — check any commercial software you rely on before ordering. NVIDIA has announced a separate DGX Station for Windows SKU (Windows plus Windows Subsystem for Linux) for a later release; the units SourceIT supplies today ship with Ubuntu.\u003c\/p\u003e\n\u003ch3\u003eClustering and Networking Compatibility\u003c\/h3\u003e\n\u003cp\u003eThe two NVIDIA ConnectX-8 SuperNIC QSFP112 ports run at 400 Gb\/s each (up to 800 Gb\/s aggregate) with RDMA over Converged Ethernet, GPUDirect RDMA and GPUDirect Storage support, so two or more systems can be linked for distributed training, or connected to an NVMe-over-fabric storage array. QSFP112 transceivers or direct-attach cables are ordered separately to match your switch fabric — SourceIT will quote compatible optics alongside the system. The 10GbE RJ45 port covers everyday site networking, and the dedicated management LAN keeps out-of-band BMC traffic isolated from production data.\u003c\/p\u003e\n\u003ch3\u003eSupermicro Enterprise Platform\u003c\/h3\u003e\n\u003cp\u003eThe Super AI Station uses the same BMC, IPMI 2.0 and Redfish management foundation as Supermicro's rack servers, so it can be monitored and serviced alongside existing Supermicro infrastructure, and the optional rail kit lets it move from deskside to a 5U rack slot without a chassis change. Supermicro also offers the system as a Gold Series ready-to-ship configuration; SourceIT quotes the configuration described on this page and can adjust storage or add a display GPU to order.\u003c\/p\u003e\n\u003c\/div\u003e\u003c\/details\u003e\n\u003cdetails class=\"pd-acc\"\u003e\u003csummary style=\"cursor: pointer; font-weight: 700;\"\u003ePackage Contents \u0026amp; Ordering Information\u003c\/summary\u003e\u003cdiv class=\"pd-acc-body\"\u003e\n\u003ch3\u003eWhat Ships With the System\u003c\/h3\u003e\n\u003cul\u003e\n\u003cli\u003eSupermicro Super AI Station ARS-511GD-NB-LCC tower with NVIDIA GB300 Grace Blackwell Ultra Desktop Superchip\u003c\/li\u003e\n\u003cli\u003e2 × 1.92 TB NVMe Gen4 enterprise M.2 SSD (1 DWPD) configured as OS RAID 1\u003c\/li\u003e\n\u003cli\u003e2 × 1.92 TB NVMe Gen4 enterprise M.2 SSD (1 DWPD) configured as data RAID 0\u003c\/li\u003e\n\u003cli\u003eUbuntu 24.04 LTS preinstalled with NVIDIA drivers, CUDA and cuDNN\u003c\/li\u003e\n\u003cli\u003e1600 W Titanium power supply (installed) and power cord\u003c\/li\u003e\n\u003cli\u003eSupermicro 3-year parts-and-service warranty with next-business-day onsite hardware support (OSNBD3)\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003cp\u003eQSFP transceivers or DAC cables, the 5U rackmount rail kit (MCP-290-75905-0B), a Wi-Fi \/ Bluetooth module and an optional NVIDIA RTX PRO Blackwell display GPU are quoted separately.\u003c\/p\u003e\n\u003ch3\u003eNVIDIA DGX Station GB300 Systems at SourceIT Singapore\u003c\/h3\u003e\n\u003cp\u003eThe Supermicro Super AI Station is quoted by SourceIT Singapore with GST invoice and Supermicro's 3-year NBD onsite warranty; a dedicated 20 A power circuit is recommended for the 1600 W platform.\u003c\/p\u003e\n\u003cul\u003e\n\u003cli\u003e\n\u003ca href=\"\/products\/asus-expertcenter-pro-et900n-g3-gb300-ai-supercomputer\"\u003eASUS ExpertCenter Pro ET900N G3\u003c\/a\u003e — NVIDIA DGX Station GB300 platform, same 748 GB \/ 20 PFLOPS core compute\u003c\/li\u003e\n\u003cli\u003e\n\u003ca href=\"\/products\/msi-xpertstation-ws300-nvidia-dgx-station-ai-supercomputer\"\u003eMSI XpertStation WS300\u003c\/a\u003e — NVIDIA DGX Station GB300 platform, same 748 GB \/ 20 PFLOPS core compute\u003c\/li\u003e\n\u003cli\u003e\n\u003ca href=\"\/products\/hp-zgx-fury-g1n-gb300-ai-workstation-with-nvidia-blackwell-ultra\"\u003eHP ZGX Fury AI Station G1n\u003c\/a\u003e — NVIDIA DGX Station GB300 platform, same 748 GB \/ 20 PFLOPS core compute\u003c\/li\u003e\n\u003cli\u003e\n\u003ca href=\"\/products\/gigabyte-w775-v10-l01-deskside-ai-supercomputer-with-nvidia-gb300-blackwell\"\u003eGIGABYTE W775-V10-L01\u003c\/a\u003e — NVIDIA DGX Station GB300 platform, same 748 GB \/ 20 PFLOPS core compute\u003c\/li\u003e\n\u003cli\u003e\n\u003ca href=\"\/products\/nvidia-dgx-spark-ai-supercomputer-4tb-940-54242-0007-000\"\u003eNVIDIA DGX Spark\u003c\/a\u003e — desktop-scale GB10 system (128 GB unified memory, 1 PFLOP FP4) for individual developers\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003cp\u003eBrowse the full \u003ca href=\"\/collections\/nvidia-dgx-spark\"\u003eNVIDIA DGX Spark \u0026amp; DGX Station collection\u003c\/a\u003e. For volume, project, research-grant or education pricing, GST invoicing and current lead times, email \u003ca href=\"mailto:sales@sourceit.com.sg\"\u003esales@sourceit.com.sg\u003c\/a\u003e.\u003c\/p\u003e\n\u003c\/div\u003e\u003c\/details\u003e\n\u003cdetails class=\"pd-acc\"\u003e\u003csummary style=\"cursor: pointer; font-weight: 700;\"\u003eDatasheet \u0026amp; Downloads\u003c\/summary\u003e\u003cdiv class=\"pd-acc-body\"\u003e\n\u003cp\u003eOfficial Supermicro documentation for the Super AI Station ARS-511GD-NB-LCC:\u003c\/p\u003e\n\u003cul\u003e\n\u003cli\u003e\u003ca href=\"https:\/\/www.supermicro.com\/datasheet\/datasheet_Supermicro_Super_AI_Station.pdf\" target=\"_blank\" rel=\"noopener\"\u003eSupermicro Super AI Station datasheet (PDF)\u003c\/a\u003e\u003c\/li\u003e\n\u003cli\u003e\u003ca href=\"https:\/\/www.supermicro.com\/en\/products\/system\/gpu\/tower%20or%205u\/ars-511gd-nb-lcc\" target=\"_blank\" rel=\"noopener\"\u003eSupermicro ARS-511GD-NB-LCC — product page\u003c\/a\u003e\u003c\/li\u003e\n\u003cli\u003e\u003ca href=\"https:\/\/www.supermicro.com\/en\/products\/system\/datasheet\/ars-511gd-nb-lcc\" target=\"_blank\" rel=\"noopener\"\u003eSupermicro ARS-511GD-NB-LCC — full specifications\u003c\/a\u003e\u003c\/li\u003e\n\u003cli\u003e\u003ca href=\"https:\/\/www.supermicro.com\/en\/accelerators\/nvidia\/super-ai-station\" target=\"_blank\" rel=\"noopener\"\u003eSupermicro Super AI Station — overview\u003c\/a\u003e\u003c\/li\u003e\n\u003cli\u003e\u003ca href=\"https:\/\/www.nvidia.com\/en-sg\/products\/workstations\/dgx-station\/\" target=\"_blank\" rel=\"noopener\"\u003eNVIDIA DGX Station (GB300) — official platform page (Singapore)\u003c\/a\u003e\u003c\/li\u003e\n\u003cli\u003e\u003ca href=\"https:\/\/www.nvidia.com\/content\/dam\/en-zz\/Solutions\/products\/workstations\/dgx-station\/workstation-datasheet-dgx-station-nvidia-us-5543600-web.pdf\" target=\"_blank\" rel=\"noopener\"\u003eNVIDIA DGX Station datasheet (PDF)\u003c\/a\u003e\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003c\/div\u003e\u003c\/details\u003e\n\u003cdetails class=\"pd-acc\"\u003e\u003csummary style=\"cursor: pointer; font-weight: 700;\"\u003eFrequently Asked Questions — Supermicro Super AI Station ARS-511GD-NB-LCC\u003c\/summary\u003e\u003cdiv class=\"pd-acc-body\"\u003e\n\u003ch3\u003eWhat is the Supermicro Super AI Station?\u003c\/h3\u003e\n\u003cp\u003eSupermicro's build of the NVIDIA DGX Station GB300 platform (model ARS-511GD-NB-LCC): a liquid-cooled tower with the GB300 Grace Blackwell Ultra Superchip, 748 GB ECC coherent memory and up to 20 PFLOPS FP4 for training, fine-tuning and serving AI models of up to 1 trillion parameters on premises — with the option to rack-mount it as a 5U unit.\u003c\/p\u003e\n\u003ch3\u003eWhat storage does it ship with?\u003c\/h3\u003e\n\u003cp\u003eFour 1.92 TB enterprise NVMe Gen4 M.2 SSDs rated at 1 DWPD: two as an OS RAID 1 pair on the PCIe 5.0 slots and two as a RAID 0 data volume on the PCIe 6.0 slots, both configured through the host OS.\u003c\/p\u003e\n\u003ch3\u003eIs the operating system preinstalled?\u003c\/h3\u003e\n\u003cp\u003eYes. Ubuntu 24.04 LTS is preinstalled with NVIDIA drivers, CUDA and cuDNN, together with the NVIDIA AI Developer Tools.\u003c\/p\u003e\n\u003ch3\u003eCan I rack-mount it?\u003c\/h3\u003e\n\u003cp\u003eYes. The chassis is a tower by default and converts to a 5U rackmount unit with Supermicro's optional MCP-290-75905-0B rail and handle kit, quoted separately.\u003c\/p\u003e\n\u003ch3\u003eCan I add a display graphics card?\u003c\/h3\u003e\n\u003cp\u003eYes. The PCIe 5.0 x16 double-width slot accepts an NVIDIA RTX PRO 2000, RTX PRO 4000 SFF, RTX PRO 6000 Max-Q or RTX PRO 6000 Workstation Edition Blackwell card — quoted separately.\u003c\/p\u003e\n\u003ch3\u003eHow does it compare to the NVIDIA DGX Spark?\u003c\/h3\u003e\n\u003cp\u003eDifferent class entirely: the Super AI Station delivers up to 20 PFLOPS FP4 and 748 GB coherent memory for models up to 1 trillion parameters, versus the DGX Spark's 1 PFLOP and 128 GB (about 200 billion parameters). Choose the Spark for individual developers; the Super AI Station for teams and production-scale AI.\u003c\/p\u003e\n\u003ch3\u003eCan multiple team members share one system?\u003c\/h3\u003e\n\u003cp\u003eYes — Multi-Instance GPU (MIG) partitions the Blackwell Ultra GPU into up to seven isolated instances for concurrent users and workloads.\u003c\/p\u003e\n\u003ch3\u003eDoes it need a server room?\u003c\/h3\u003e\n\u003cp\u003eNo rack is required — cooling is a closed-loop direct-to-chip liquid loop inside the chassis. Plan an air-conditioned space within Supermicro's 15–32 °C operating range and a dedicated 20 A power circuit for the 1600 W Titanium power supply.\u003c\/p\u003e\n\u003ch3\u003eWhat warranty does it come with?\u003c\/h3\u003e\n\u003cp\u003eSupermicro's 3-year parts-and-service warranty with next-business-day onsite hardware support (OSNBD3), supplied through SourceIT with local support in Singapore.\u003c\/p\u003e\n\u003ch3\u003eHow do I purchase the Supermicro Super AI Station in Singapore?\u003c\/h3\u003e\n\u003cp\u003eSourceIT is an authorised IT reseller in Singapore and supplies the ARS-511GD-NB-LCC with GST invoice, Supermicro onsite warranty and enterprise procurement assistance — contact sales@sourceit.com.sg for configuration options, volume pricing and lead times.\u003c\/p\u003e\n\u003c\/div\u003e\u003c\/details\u003e\n\u003ch3\u003eProduct Video\u003c\/h3\u003e\n\u003cp\u003e\u003ciframe width=\"560\" height=\"315\" src=\"https:\/\/www.youtube.com\/embed\/7YhgDWpedJA\" title=\"The Future of AI Development Solutions - Super AI Station | Supermicro\" frameborder=\"0\" allow=\"accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture\" allowfullscreen\u003e\u003c\/iframe\u003e\u003c\/p\u003e\n\u003c\/div\u003e","brand":"Supermicro","offers":[{"title":"Default Title","offer_id":51487892799652,"sku":"ARS-511GD-NB-LCC","price":150000.0,"currency_code":"SGD","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0605\/0136\/0804\/files\/supermicro-ars-511gd-nb-lcc-dgx-station-gb300-1.png?v=1790296850"},{"product_id":"dell-pro-max-with-gb300-nvidia-dgx-station-ai-workstation-fct6263","title":"Dell Pro Max with GB300 NVIDIA DGX Station AI Workstation (FCT6263)","description":"\u003cdiv\u003e\n\u003cstyle\u003e\n.pd-acc { border: 1px solid #e2e2e2; border-radius: 8px; margin: 10px 0; overflow: hidden; }\n.pd-acc summary { cursor: pointer; padding: 14px 16px; font-weight: 700; font-size: 1.05em; list-style: none; display: flex; justify-content: space-between; align-items: center; background: #f7f7f7; }\n.pd-acc[open] \u003e summary { border-bottom: 1px solid #e2e2e2; }\n.pd-acc summary::-webkit-details-marker { display: none; }\n.pd-acc summary::after { content: \"+\"; font-size: 1.4em; font-weight: 400; line-height: 1; margin-left: 12px; }\n.pd-acc[open] \u003e summary::after { content: \"\\2212\"; }\n.pd-acc .pd-acc-body { padding: 8px 16px 14px; }\n.pd-spec, .pd-spec tr, .pd-spec td { border: none !important; background: transparent !important; }\n.pd-spec { width: 100%; border-collapse: collapse !important; margin: 6px 0 18px; }\n.pd-spec td { padding: 7px 10px 7px 0 !important; border-bottom: 1px solid #ececec !important; vertical-align: top; font-size: 0.95em; text-align: left; }\n.pd-spec td:first-child { width: 38%; font-weight: 600; color: #333; }\n.pd-spec tr:last-child td { border-bottom: none !important; }\n\u003c\/style\u003e\n\u003ch2\u003eDell Pro Max with GB300 NVIDIA DGX Station Grace Blackwell Ultra Deskside AI Supercomputer, 748GB Coherent Memory, 20 PFLOPS, RTX PRO 2000 Blackwell, 8TB or 16TB NVMe, Dual 400G ConnectX-8, MaxCool Liquid Cooling (FCT6263) - 3 Years Dell ProSupport\u003c\/h2\u003e\n\u003ch3\u003eDell Pro Max with GB300 — The AI Supercomputer Built for Your Desk\u003c\/h3\u003e\n\u003cp\u003eThe Dell Pro Max with GB300 (model FCT6263, now also sold by Dell as the Dell Pro Precision with GB300) is Dell's build of the NVIDIA DGX Station GB300 platform — and Dell was the first OEM to ship a GB300 desktop. The NVIDIA GB300 Grace Blackwell Ultra Desktop Superchip pairs a 72-core NVIDIA Grace CPU (Arm Neoverse V2) with an NVIDIA Blackwell Ultra GPU over NVLink-C2C, sharing 748 GB of coherent memory and delivering up to 20 petaFLOPS of FP4 AI performance for prototyping, fine-tuning and serving AI models of up to 1 trillion parameters locally — with no cloud dependency.\u003c\/p\u003e\n\u003cp\u003eFor Singapore enterprises, research institutes, universities and AI teams, it brings data-centre-class compute under your own roof and your own data governance, ready for agentic AI, long-context LLM post-training, RAG, data science and edge AI development.\u003c\/p\u003e\n\u003ch3\u003e748 GB Coherent Memory for Trillion-Parameter Models\u003c\/h3\u003e\n\u003cp\u003e496 GB of LPDDR5X CPU memory and 252 GB of HBM3e GPU memory are unified over NVLink-C2C into a single 748 GB coherent memory space, so frontier-scale models and large datasets stay resident without CPU–GPU offloading. Multi-Instance GPU (MIG) partitions the Blackwell Ultra GPU into up to seven isolated instances, giving up to seven users their own protected slice of one system.\u003c\/p\u003e\n\u003ch3\u003eExclusive Dell MaxCool Technology\u003c\/h3\u003e\n\u003cp\u003eDell's MaxCool parallel liquid-cooling architecture — smart cold-plate design, dual heat exchangers and next-generation thermal interface materials, with redundant pumps — removes heat up to 5x more efficiently, per Dell, so the Superchip holds full performance through sustained 24 × 7 training and inference without thermal throttling.\u003c\/p\u003e\n\u003ch3\u003eChoose 8 TB or 16 TB of SED NVMe, Plus an RTX PRO 2000 Blackwell Display GPU\u003c\/h3\u003e\n\u003cp\u003eSelect your configuration above, exactly as Dell offers it: 16 TB (4 × 4 TB M.2 2280 PCIe Gen4 x4 NVMe TLC SED) or 8 TB (4 × 2 TB M.2 2280 PCIe Gen5 x4 NVMe TLC SED), both configured as RAID 1 on SSD1 and SSD2 with SSD3 and SSD4 as individual drives. Every unit includes an NVIDIA RTX PRO 2000 Blackwell Workstation Edition (75 W, 16 GB GDDR7, 4 × mini-DisplayPort) for local displays, and ships with or without the Dell KB216 wired keyboard (Dell MS116 wired mouse included).\u003c\/p\u003e\n\u003ch3\u003eDual 400 Gb\/s ConnectX-8 SuperNIC Networking\u003c\/h3\u003e\n\u003cp\u003eTwo QSFP112 400 Gb\/s ports on the NVIDIA ConnectX-8 SuperNIC (up to 800 Gb\/s) link two Dell Pro Max with GB300 systems to double performance for larger models, or connect into private AI clusters and high-speed storage, alongside 10 GbE and 1 GbE RJ45 ports for site and management networking.\u003c\/p\u003e\n\u003ch3\u003eUbuntu 24.04 LTS with NVIDIA AI Developer Tools and Ubuntu Pro\u003c\/h3\u003e\n\u003cp\u003eThe system ships with Ubuntu 24.04 LTS, the NVIDIA AI Developer Tools and Ubuntu Pro, ready for CUDA, NVIDIA NeMo, NIM, TensorRT-LLM, vLLM and NVIDIA OpenShell \/ NemoClaw agent workflows on delivery.\u003c\/p\u003e\n\u003ch3\u003eAvailable in Singapore Through SourceIT\u003c\/h3\u003e\n\u003cp\u003eSourceIT is an authorised IT reseller in Singapore and supplies the Dell Pro Max with GB300 with a full GST invoice, Dell's 3-year ProSupport with onsite service after remote diagnosis, and enterprise procurement support — email \u003ca href=\"mailto:sales@sourceit.com.sg\"\u003esales@sourceit.com.sg\u003c\/a\u003e or use the Request for Quotation button for pricing, lead time and configuration advice.\u003c\/p\u003e\n\u003cdetails class=\"pd-acc\"\u003e\u003csummary style=\"cursor: pointer; font-weight: 700;\"\u003eDell Pro Max with GB300 Datasheet — Technical Specifications\u003c\/summary\u003e\u003cdiv class=\"pd-acc-body\"\u003e\n\u003ch3\u003eGeneral\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eProduct Name\u003c\/td\u003e\n\u003ctd\u003eDell Pro Max with GB300 (Dell Pro Precision with GB300)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eModel\u003c\/td\u003e\n\u003ctd\u003eFCT6263\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eCategory\u003c\/td\u003e\n\u003ctd\u003eDeskside AI supercomputer \/ AI workstation (NVIDIA DGX Station GB300 platform)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eBase Chassis\u003c\/td\u003e\n\u003ctd\u003eDell Pro Precision with GB300 Chassis MP, V2\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eWarranty\u003c\/td\u003e\n\u003ctd\u003eProSupport with Onsite Service After Remote Diagnosis, 36 months (base: Basic Hardware Service with Onsite Support After Remote Diagnosis, 12 months)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eLocal Warranty (Singapore)\u003c\/td\u003e\n\u003ctd\u003e3 Years Dell ProSupport; procurement and GST invoice by SourceIT\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\u003ch3\u003eProcessor \u0026amp; GPU\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eSuperchip\u003c\/td\u003e\n\u003ctd\u003eNVIDIA GB300 Grace Blackwell Ultra Desktop Superchip\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eCPU\u003c\/td\u003e\n\u003ctd\u003eNVIDIA Grace, 72 Arm Neoverse V2 cores, with ConnectX-8 SuperNIC\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eGPU\u003c\/td\u003e\n\u003ctd\u003eNVIDIA Blackwell Ultra GPU\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eCPU–GPU Interconnect\u003c\/td\u003e\n\u003ctd\u003eNVIDIA NVLink-C2C, 900 GB\/s (per NVIDIA DGX Station datasheet)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eAI Performance\u003c\/td\u003e\n\u003ctd\u003eUp to 20 PFLOPS FP4 Tensor Core (with sparsity), per NVIDIA and Dell\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eModel Support\u003c\/td\u003e\n\u003ctd\u003eAI models up to 1 trillion parameters\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eGPU Partitioning\u003c\/td\u003e\n\u003ctd\u003eMulti-Instance GPU (MIG) — up to 7 instances\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eDisplay Graphics\u003c\/td\u003e\n\u003ctd\u003eNVIDIA RTX PRO 2000 Blackwell Workstation Edition (75 W), 16 GB GDDR7, 4 × mDP to DP\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\u003ch3\u003eMemory\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eTotal Coherent Memory\u003c\/td\u003e\n\u003ctd\u003e748 GB unified CPU + GPU memory\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eCPU Memory\u003c\/td\u003e\n\u003ctd\u003e496 GB LPDDR5X\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eGPU Memory\u003c\/td\u003e\n\u003ctd\u003e252 GB HBM3e\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\u003ch3\u003eStorage (select configuration)\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003e16 TB option\u003c\/td\u003e\n\u003ctd\u003e4 × 4 TB M.2 2280 SSD with DRAM, PCIe Gen4 x4 NVMe TLC SED\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003e8 TB option\u003c\/td\u003e\n\u003ctd\u003e4 × 2 TB M.2 2280 SSD with DRAM, PCIe Gen5 x4 NVMe TLC SED\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eRAID\u003c\/td\u003e\n\u003ctd\u003eRAID 1 for SSD1 and SSD2, no RAID for SSD3 and SSD4\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\u003ch3\u003ePorts \u0026amp; Slots\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eTop Ports\u003c\/td\u003e\n\u003ctd\u003e2 × USB 3.2 Gen 2 (10 Gbps), 2 × USB 3.2 Gen 2 (10 Gbps) Type-C\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eRear Ports\u003c\/td\u003e\n\u003ctd\u003e2 × QSFP112 (400 Gbps), 4 × USB 3.2 Gen 2 (10 Gbps), 1 × RJ45 1 GbE, 1 × RJ45 10 GbE, line-out, line-in, audio input\/microphone\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eSlots\u003c\/td\u003e\n\u003ctd\u003e1 × Kensington security-cable slot\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\u003ch3\u003eSoftware \u0026amp; Peripherals\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eOperating System\u003c\/td\u003e\n\u003ctd\u003eUbuntu 24.04 LTS with NVIDIA AI Developer Tools with Ubuntu Pro\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eKeyboard\u003c\/td\u003e\n\u003ctd\u003eDell Pro 3 Wired Keyboard KB216, English-US, Black (or no keyboard — select above)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eMouse\u003c\/td\u003e\n\u003ctd\u003eDell Pro 3 Wired Mouse MS116, Black\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\u003ch3\u003ePower, Cooling \u0026amp; Physical\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eCooling\u003c\/td\u003e\n\u003ctd\u003eDell MaxCool parallel liquid cooling with redundant pumps\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003ePower Cord\u003c\/td\u003e\n\u003ctd\u003eSystem Power Cord, C19, 2.44 m, 250 V 13 A, UK\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eDimensions (H × W × D)\u003c\/td\u003e\n\u003ctd\u003e569.00 × 231.60 × 610.50 mm (22.40 × 9.12 × 24.04 in)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eWeight\u003c\/td\u003e\n\u003ctd\u003eUp to 38.67 kg (85.25 lb)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eChassis\u003c\/td\u003e\n\u003ctd\u003eBlack powder-coated metal main chassis; Magnetite painted metal doors; Magnetite painted resin top, front and bottom covers\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\u003c\/div\u003e\u003c\/details\u003e\n\u003cdetails class=\"pd-acc\"\u003e\u003csummary style=\"cursor: pointer; font-weight: 700;\"\u003eCompatibility \u0026amp; Software Support\u003c\/summary\u003e\u003cdiv class=\"pd-acc-body\"\u003e\n\u003ch3\u003eNVIDIA AI Software Stack\u003c\/h3\u003e\n\u003cp\u003eUbuntu 24.04 LTS with the NVIDIA AI Developer Tools — the same NVIDIA software foundation used across the DGX family. CUDA, cuDNN, TensorRT and TensorRT-LLM, NVIDIA NeMo, NIM microservices, Triton, RAPIDS and NGC containers run natively, alongside PyTorch, JAX, vLLM and Hugging Face Transformers; Dell supports NVIDIA OpenShell and NemoClaw for secure, long-running local agents. NVIDIA AI Enterprise is available as a separate subscription.\u003c\/p\u003e\n\u003ch3\u003eArm64 Architecture — Plan Your Software Accordingly\u003c\/h3\u003e\n\u003cp\u003eThe Grace CPU is Arm Neoverse V2, so the OS, drivers and containers are arm64 (aarch64) builds. Mainstream AI frameworks and NGC containers are published for arm64, but x86-only proprietary applications will not run natively — check commercial software before ordering. Dell has announced a Windows-based Dell Pro Precision with GB300; units on this listing ship with Ubuntu.\u003c\/p\u003e\n\u003ch3\u003eClustering and Networking\u003c\/h3\u003e\n\u003cp\u003eThe two ConnectX-8 QSFP112 400 Gb\/s ports link two systems for larger models or connect to NVMe-over-fabric storage. QSFP112 transceivers or DAC cables are ordered separately — SourceIT quotes compatible optics alongside the system.\u003c\/p\u003e\n\u003c\/div\u003e\u003c\/details\u003e\n\u003cdetails class=\"pd-acc\"\u003e\u003csummary style=\"cursor: pointer; font-weight: 700;\"\u003ePackage Contents \u0026amp; Ordering Information\u003c\/summary\u003e\u003cdiv class=\"pd-acc-body\"\u003e\n\u003ch3\u003eWhat Ships With the System\u003c\/h3\u003e\n\u003cul\u003e\n\u003cli\u003eDell Pro Max with GB300 (FCT6263) with NVIDIA GB300 Grace Blackwell Ultra Desktop Superchip and RTX PRO 2000 Blackwell\u003c\/li\u003e\n\u003cli\u003e8 TB or 16 TB SED NVMe storage as selected\u003c\/li\u003e\n\u003cli\u003eDell KB216 wired keyboard (if selected) and MS116 wired mouse\u003c\/li\u003e\n\u003cli\u003eC19 system power cord (2.44 m, UK) and Dell documentation\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003eWhich Configuration?\u003c\/h3\u003e\n\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003e16 TB (4 × 4 TB PCIe Gen4)\u003c\/strong\u003e — Dell's default build, maximum on-board dataset and checkpoint capacity\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003e8 TB (4 × 2 TB PCIe Gen5)\u003c\/strong\u003e — faster Gen5 drives at lower capacity\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003eNVIDIA DGX Station GB300 Systems at SourceIT Singapore\u003c\/h3\u003e\n\u003cul\u003e\n\u003cli\u003e\n\u003ca href=\"\/products\/hp-zgx-fury-g1n-gb300-ai-workstation-with-nvidia-blackwell-ultra\"\u003eHP ZGX Fury AI Station G1n\u003c\/a\u003e — NVIDIA DGX Station GB300 platform\u003c\/li\u003e\n\u003cli\u003e\n\u003ca href=\"\/products\/asus-expertcenter-pro-et900n-g3-gb300-ai-supercomputer\"\u003eASUS ExpertCenter Pro ET900N G3\u003c\/a\u003e — NVIDIA DGX Station GB300 platform\u003c\/li\u003e\n\u003cli\u003e\n\u003ca href=\"\/products\/msi-xpertstation-ws300-nvidia-dgx-station-ai-supercomputer\"\u003eMSI XpertStation WS300\u003c\/a\u003e — NVIDIA DGX Station GB300 platform\u003c\/li\u003e\n\u003cli\u003e\n\u003ca href=\"\/products\/gigabyte-w775-v10-l01-deskside-ai-supercomputer-with-nvidia-gb300-blackwell\"\u003eGIGABYTE W775-V10-L01\u003c\/a\u003e — NVIDIA DGX Station GB300 platform\u003c\/li\u003e\n\u003cli\u003e\n\u003ca href=\"\/products\/supermicro-super-ai-station-ars-511gd-nb-lcc-nvidia-dgx-station-gb300\"\u003eSupermicro Super AI Station ARS-511GD-NB-LCC\u003c\/a\u003e — NVIDIA DGX Station GB300 platform\u003c\/li\u003e\n\u003cli\u003e\n\u003ca href=\"\/products\/dell-pro-max-ai-desktop-pcs-with-nvidia-gb10-blackwell-gpu-4tb\"\u003eDell Pro Max with GB10\u003c\/a\u003e — desktop-scale GB10 system (128 GB unified memory, 1 PFLOP FP4) for individual developers\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003cp\u003eBrowse the full \u003ca href=\"\/collections\/nvidia-dgx-spark\"\u003eNVIDIA DGX Spark \u0026amp; DGX Station collection\u003c\/a\u003e. For project, research-grant or education pricing, GST invoicing and lead times, email \u003ca href=\"mailto:sales@sourceit.com.sg\"\u003esales@sourceit.com.sg\u003c\/a\u003e quoting FCT6263.\u003c\/p\u003e\n\u003c\/div\u003e\u003c\/details\u003e\n\u003cdetails class=\"pd-acc\"\u003e\u003csummary style=\"cursor: pointer; font-weight: 700;\"\u003eDatasheet \u0026amp; Downloads\u003c\/summary\u003e\u003cdiv class=\"pd-acc-body\"\u003e\n\u003cul\u003e\n\u003cli\u003e\u003ca href=\"https:\/\/www.dell.com\/en-sg\/shop\/desktop-computers\/spd\/dellpromaxwithgb300fct6263\" target=\"_blank\" rel=\"noopener\"\u003eDell Pro Max with GB300 — Dell Singapore product page\u003c\/a\u003e\u003c\/li\u003e\n\u003cli\u003e\u003ca href=\"https:\/\/www.dell.com\/support\/product-details\/en-sg\/product\/dellpromaxwithgb300fct6263\/overview\" target=\"_blank\" rel=\"noopener\"\u003eDell Pro Max with GB300 — drivers, manuals \u0026amp; support\u003c\/a\u003e\u003c\/li\u003e\n\u003cli\u003e\u003ca href=\"https:\/\/www.nvidia.com\/en-sg\/products\/workstations\/dgx-station\/\" target=\"_blank\" rel=\"noopener\"\u003eNVIDIA DGX Station (GB300) — official platform page (Singapore)\u003c\/a\u003e\u003c\/li\u003e\n\u003cli\u003e\u003ca href=\"https:\/\/www.nvidia.com\/content\/dam\/en-zz\/Solutions\/products\/workstations\/dgx-station\/workstation-datasheet-dgx-station-nvidia-us-5543600-web.pdf\" target=\"_blank\" rel=\"noopener\"\u003eNVIDIA DGX Station datasheet (PDF)\u003c\/a\u003e\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003c\/div\u003e\u003c\/details\u003e\n\u003cdetails class=\"pd-acc\"\u003e\u003csummary style=\"cursor: pointer; font-weight: 700;\"\u003eFrequently Asked Questions — Dell Pro Max with GB300\u003c\/summary\u003e\u003cdiv class=\"pd-acc-body\"\u003e\n\u003ch3\u003eWhat is the Dell Pro Max with GB300?\u003c\/h3\u003e\n\u003cp\u003eDell's deskside AI supercomputer on the NVIDIA DGX Station GB300 platform: a 72-core Grace CPU and Blackwell Ultra GPU sharing 748 GB of coherent memory, up to 20 PFLOPS FP4, cooled by Dell MaxCool — for running and fine-tuning AI models of up to 1 trillion parameters locally.\u003c\/p\u003e\n\u003ch3\u003eIs it the same as the Dell Pro Precision with GB300?\u003c\/h3\u003e\n\u003cp\u003eYes — Dell's Singapore store now lists model FCT6263 as the Dell Pro Precision with GB300. It is the same system.\u003c\/p\u003e\n\u003ch3\u003eWhat is the difference between the configurations?\u003c\/h3\u003e\n\u003cp\u003eStorage only, plus the optional keyboard: 16 TB uses 4 × 4 TB PCIe Gen4 SED NVMe, 8 TB uses 4 × 2 TB PCIe Gen5 SED NVMe. Superchip, memory, RTX PRO 2000 Blackwell, networking and warranty are identical.\u003c\/p\u003e\n\u003ch3\u003eHow does it compare to the Dell Pro Max with GB10?\u003c\/h3\u003e\n\u003cp\u003eThe GB10 model is a desktop-scale DGX Spark-class system with 128 GB unified memory and 1 PFLOP FP4 for individual developers; the GB300 delivers 748 GB and up to 20 PFLOPS for teams and trillion-parameter-scale AI.\u003c\/p\u003e\n\u003ch3\u003eCan several users share one system?\u003c\/h3\u003e\n\u003cp\u003eYes — MIG partitions the Blackwell Ultra GPU into up to seven isolated instances.\u003c\/p\u003e\n\u003ch3\u003eDoes it run Windows?\u003c\/h3\u003e\n\u003cp\u003eThis configuration ships with Ubuntu 24.04 LTS. Dell has announced a Windows version of the Dell Pro Precision with GB300; ask us about availability.\u003c\/p\u003e\n\u003ch3\u003eWhat warranty is included?\u003c\/h3\u003e\n\u003cp\u003eDell ProSupport with Onsite Service After Remote Diagnosis for 36 months, as configured on Dell Singapore.\u003c\/p\u003e\n\u003ch3\u003eHow do I buy it in Singapore?\u003c\/h3\u003e\n\u003cp\u003eSourceIT is an authorised IT reseller in Singapore — email \u003ca href=\"mailto:sales@sourceit.com.sg\"\u003esales@sourceit.com.sg\u003c\/a\u003e or use the Request for Quotation button for the latest price, availability and lead time.\u003c\/p\u003e\n\u003c\/div\u003e\u003c\/details\u003e\n\u003ch3\u003eProduct Video\u003c\/h3\u003e\n\u003cp\u003e\u003ciframe width=\"560\" height=\"315\" src=\"https:\/\/www.youtube.com\/embed\/QgfH3-i03is\" title=\"Dell Pro Max with GB300 and NVIDIA NemoClaw | Dell Technologies\" frameborder=\"0\" allow=\"accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture\" allowfullscreen\u003e\u003c\/iframe\u003e\u003c\/p\u003e\n\u003c\/div\u003e\n","brand":"Dell","offers":[{"title":"16TB (4x4TB PCIe Gen4 SED) \/ Dell KB216 Wired Keyboard","offer_id":57608988065956,"sku":"FCT6263-16TB-KB","price":170000.0,"currency_code":"SGD","in_stock":true},{"title":"16TB (4x4TB PCIe Gen4 SED) \/ No Keyboard","offer_id":57608988098724,"sku":"FCT6263-16TB-NOKB","price":169990.19,"currency_code":"SGD","in_stock":true},{"title":"8TB (4x2TB PCIe Gen5 SED) \/ Dell KB216 Wired Keyboard","offer_id":57608988131492,"sku":"FCT6263-8TB-KB","price":159309.28,"currency_code":"SGD","in_stock":true},{"title":"8TB (4x2TB PCIe Gen5 SED) \/ No Keyboard","offer_id":57608988164260,"sku":"FCT6263-8TB-NOKB","price":159299.47,"currency_code":"SGD","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0605\/0136\/0804\/files\/desktop-dell-pro-max-fct6263-bk-gallery-1.jpg?v=1793847268"},{"product_id":"lenovo-nvidia-gb300-nvl72-rack-scale-ai-supercomputer-72-blackwell-ultra-gpus-36-grace-cpus-liquid-cooled","title":"Lenovo NVIDIA GB300 NVL72 Rack Scale AI Supercomputer - 72 Blackwell Ultra GPUs, 36 Grace CPUs, Liquid Cooled","description":"\u003cdiv\u003e\n\u003cstyle\u003e\n.pd-acc { border: 1px solid #e2e2e2; border-radius: 8px; margin: 10px 0; overflow: hidden; }\n.pd-acc summary { cursor: pointer; padding: 14px 16px; font-weight: 700; font-size: 1.05em; list-style: none; display: flex; justify-content: space-between; align-items: center; background: #f7f7f7; }\n.pd-acc[open] \u003e summary { border-bottom: 1px solid #e2e2e2; }\n.pd-acc summary::-webkit-details-marker { display: none; }\n.pd-acc summary::after { content: \"+\"; font-size: 1.4em; font-weight: 400; line-height: 1; margin-left: 12px; }\n.pd-acc[open] \u003e summary::after { content: \"\\2212\"; }\n.pd-acc .pd-acc-body { padding: 8px 16px 14px; }\n.pd-spec, .pd-spec tr, .pd-spec td { border: none !important; background: transparent !important; }\n.pd-spec { width: 100%; border-collapse: collapse !important; margin: 6px 0 18px; }\n.pd-spec td { padding: 7px 10px 7px 0 !important; border-bottom: 1px solid #ececec !important; vertical-align: top; font-size: 0.95em; text-align: left; }\n.pd-spec td:first-child { width: 38%; font-weight: 600; color: #333; }\n.pd-spec tr:last-child td { border-bottom: none !important; }\n\u003c\/style\u003e\n\u003ch2\u003eLenovo NVIDIA GB300 NVL72 Rack Scale AI Supercomputer, 72 NVIDIA Blackwell Ultra GPUs, 36 NVIDIA Grace CPUs, 5th-Gen NVLink, 48U MGX Rack, Lenovo Neptune Liquid Cooling (7DJVCTO2WW)\u003c\/h2\u003e\n\u003ch3\u003eA Complete AI Factory Rack: 72 Blackwell Ultra GPUs in One NVLink Domain\u003c\/h3\u003e\n\u003cp\u003eThe Lenovo NVIDIA GB300 NVL72 is a fully integrated, liquid-cooled 48U rack built on the NVIDIA GB300 NVL72 platform. Eighteen 1U compute trays carry 72 NVIDIA Blackwell Ultra (B300) GPUs and 36 NVIDIA Grace CPUs, joined by nine NVLink switch trays into a single fifth-generation NVLink domain, so the whole rack behaves like one very large GPU for trillion-parameter LLM training, multimodal reasoning and high-throughput inference.\u003c\/p\u003e\n\u003cp\u003eSourceIT supplies the Lenovo GB300 NVL72 in Singapore as a project sale: site survey, power and cooling readiness, configuration, delivery and Lenovo services are scoped and quoted per deployment.\u003c\/p\u003e\n\u003ch3\u003e20.7 TB of GPU Memory and 17 TB of CPU Memory per Rack\u003c\/h3\u003e\n\u003cp\u003eEach Blackwell Ultra GPU carries 288 GB of HBM3e, and each compute tray adds 960 GB of LPDDR5X ECC memory across its two Grace CPUs. Across 18 trays that is about 20.7 TB of GPU memory and 17 TB of CPU memory. GPUs talk to each other over NVLink at 1.8 TB\/s in a full mesh.\u003c\/p\u003e\n\u003ch3\u003eHybrid Liquid Cooling Based on Lenovo Neptune\u003c\/h3\u003e\n\u003cp\u003eCPUs, GPUs and HBM are direct liquid cooled, with roughly 90% of the rack's heat removed by liquid and 10% by air. The rack accepts inlet water up to 45 °C, which can reduce or remove the need for chillers. Manifolds are available in top-feed or bottom-feed versions to suit the data hall.\u003c\/p\u003e\n\u003ch3\u003eConnectX-8 800 Gb\/s Networking and BlueField-3 DPUs\u003c\/h3\u003e\n\u003cp\u003eEvery compute tray provides four OSFP ports on NVIDIA ConnectX-8 at up to 800 Gb\/s each, plus PCIe Gen5 slots for NVIDIA BlueField-3 DPUs, for scale-out to multi-rack AI clusters.\u003c\/p\u003e\n\u003ch3\u003eFactory Integrated, Tested at Rack Level\u003c\/h3\u003e\n\u003cp\u003eLenovo builds, cables, fills and retests the system at rack level before shipment. The design targets 24x7 operation with toolless tray covers and non-disruptive firmware updates.\u003c\/p\u003e\n\u003cdetails class=\"pd-acc\"\u003e\u003csummary style=\"cursor: pointer; font-weight: 700;\"\u003eLenovo NVIDIA GB300 NVL72 Datasheet - Technical Specifications\u003c\/summary\u003e\u003cdiv class=\"pd-acc-body\"\u003e\n\u003ch3\u003eGeneral\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eProduct Name\u003c\/td\u003e\n\u003ctd\u003eLenovo NVIDIA GB300 NVL72 Rack Scale AI\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eBase Models (CTO)\u003c\/td\u003e\n\u003ctd\u003eRack 7DJVCTO2WW; compute tray 7DLZCTO2WW; NVL switch tray 7DJYCTO3WW; SN2201 management switch 7DJWCTO1WW\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003ePlatform\u003c\/td\u003e\n\u003ctd\u003eNVIDIA GB300 NVL72 (Grace Blackwell Ultra), NVIDIA MGX rack compliant\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eForm Factor\u003c\/td\u003e\n\u003ctd\u003e48U MGX rack\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eWarranty\u003c\/td\u003e\n\u003ctd\u003e3-year (5-year optional) customer-replaceable unit and onsite limited warranty, 9x5 next business day; upgrades available through Lenovo Services\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\u003ch3\u003eRack Configuration\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eCompute Trays\u003c\/td\u003e\n\u003ctd\u003e18 x 1U GB300 compute trays (2 x GB300 HPM each)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eNVLink Switch Trays\u003c\/td\u003e\n\u003ctd\u003e9 x 1U NVL switch trays\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eGPUs per Rack\u003c\/td\u003e\n\u003ctd\u003e72 x NVIDIA Blackwell Ultra (B300)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eCPUs per Rack\u003c\/td\u003e\n\u003ctd\u003e36 x NVIDIA Grace (72 Arm Neoverse V2 cores each)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003ePower Shelves\u003c\/td\u003e\n\u003ctd\u003e6 x 33 kW (non-redundant) or 8 x 33 kW (redundant), 1U each\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eTop-of-Rack Switches\u003c\/td\u003e\n\u003ctd\u003e2 x NVIDIA SN2201 (1GbE \/ 100GbE management)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eOther Rack Components\u003c\/td\u003e\n\u003ctd\u003e4 cable cartridge backplanes, 1 CDU, 2 liquid manifolds, 1400 A busbar\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\u003ch3\u003eCompute Tray (per tray)\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eProcessors\u003c\/td\u003e\n\u003ctd\u003e2 x NVIDIA Grace, 72 cores each, 300 W TDP\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eGPUs\u003c\/td\u003e\n\u003ctd\u003e4 x NVIDIA Blackwell Ultra B300, 288 GB HBM3e each\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eCPU Memory\u003c\/td\u003e\n\u003ctd\u003e960 GB LPDDR5X ECC (480 GB per CPU)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eStorage\u003c\/td\u003e\n\u003ctd\u003eUp to 8 x E1.S NVMe (PCIe Gen5), 4 by default; 1 x 1.92 TB M.2 NVMe boot; software RAID\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eNetworking\u003c\/td\u003e\n\u003ctd\u003e4 x OSFP (NVIDIA ConnectX-8), up to 800 Gb\/s each; 3 x PCIe Gen5 x16 slots for BlueField-3 DPU and OSFP boards\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eManagement\u003c\/td\u003e\n\u003ctd\u003eASPEED AST2600 BMC, 1GbE RJ-45 management port, IPMI 2.0, Redfish\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eSecurity\u003c\/td\u003e\n\u003ctd\u003eTPM 2.0, Secure Boot\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eDimensions \/ Weight\u003c\/td\u003e\n\u003ctd\u003e438 x 43.6 x 766 mm (W x H x D); 29 kg\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eOperating System\u003c\/td\u003e\n\u003ctd\u003eUbuntu (supported and certified)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\u003ch3\u003eGPU and Interconnect (per GPU)\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eGPU Memory\u003c\/td\u003e\n\u003ctd\u003e288 GB HBM3e, ECC\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eGPU-to-GPU NVLink\u003c\/td\u003e\n\u003ctd\u003e1.8 TB\/s, full mesh across the rack\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eMulti-Instance GPU\u003c\/td\u003e\n\u003ctd\u003eUp to 7 instances\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eGPU Power\u003c\/td\u003e\n\u003ctd\u003e1100 W TGP\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\u003ch3\u003ePower and Cooling\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eRack Power\u003c\/td\u003e\n\u003ctd\u003e135 kW TDP, up to 155 kW peak\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003ePower Shelf\u003c\/td\u003e\n\u003ctd\u003e33 kW, 6 x 5.5 kW hot-swap PSUs, 50 V DC output; input 200-277 V AC single or three-phase (346-480 V AC WYE)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eCooling\u003c\/td\u003e\n\u003ctd\u003eHybrid: about 90% liquid, 10% air; compute tray fans for PDB and M.2; NVL switch trays fully direct water cooled\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eCoolant\u003c\/td\u003e\n\u003ctd\u003eDeionized water (recommended) or PG25; inlet up to 45 °C\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eManifold\u003c\/td\u003e\n\u003ctd\u003eTop feed or bottom feed, stainless steel\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eOperating Air Temperature\u003c\/td\u003e\n\u003ctd\u003e10 °C to 35 °C (ASHRAE A2)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\u003c\/div\u003e\u003c\/details\u003e\n\u003cdetails class=\"pd-acc\"\u003e\u003csummary style=\"cursor: pointer; font-weight: 700;\"\u003eCompatibility \u0026amp; Software Support\u003c\/summary\u003e\u003cdiv class=\"pd-acc-body\"\u003e\n\u003ch3\u003eNVIDIA AI Software\u003c\/h3\u003e\n\u003cp\u003eThe rack runs the NVIDIA accelerated computing stack: CUDA, NVIDIA AI Enterprise, NIM microservices and NGC containers on Ubuntu for arm64. NVIDIA management software provides power monitoring and leak detection.\u003c\/p\u003e\n\u003ch3\u003eLenovo Cluster Management\u003c\/h3\u003e\n\u003cp\u003eLenovo HPC \u0026amp; AI Software Stack and Confluent cluster management are supported, with out-of-band management through BMC and HMC over IPMI 2.0 and Redfish.\u003c\/p\u003e\n\u003ch3\u003eArm64 Architecture\u003c\/h3\u003e\n\u003cp\u003eThe Grace CPU is an Arm Neoverse V2 processor, so operating system and container images are arm64 builds. Check any x86-only commercial software before ordering.\u003c\/p\u003e\n\u003ch3\u003eScale-Out Networking\u003c\/h3\u003e\n\u003cp\u003eConnectX-8 OSFP ports connect to NVIDIA Quantum-X800 InfiniBand or Spectrum-X Ethernet fabrics. Switches, optics and cabling for the cluster fabric are scoped with the project.\u003c\/p\u003e\n\u003c\/div\u003e\u003c\/details\u003e\n\u003cdetails class=\"pd-acc\"\u003e\u003csummary style=\"cursor: pointer; font-weight: 700;\"\u003eSite Requirements \u0026amp; Ordering Information\u003c\/summary\u003e\u003cdiv class=\"pd-acc-body\"\u003e\n\u003ch3\u003eWhat to Plan For\u003c\/h3\u003e\n\u003cul\u003e\n\u003cli\u003eUp to 155 kW peak per rack on three-phase power\u003c\/li\u003e\n\u003cli\u003eFacility water loop or CDU capacity for direct liquid cooling, inlet up to 45 °C\u003c\/li\u003e\n\u003cli\u003eFloor loading, delivery path and lift access for a fully populated 48U rack\u003c\/li\u003e\n\u003cli\u003eCluster fabric, storage and management networks\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003eHow This Is Sold\u003c\/h3\u003e\n\u003cp\u003eThe price shown is an indicative price for one rack. Final pricing depends on configuration (power redundancy, storage, networking, services) and is confirmed by formal quotation. The rack ships without coolant. Email \u003ca href=\"mailto:sales@sourceit.com.sg\"\u003esales@sourceit.com.sg\u003c\/a\u003e for configuration, lead time and a GST quotation.\u003c\/p\u003e\n\u003cp\u003eFor deskside AI systems, see the \u003ca href=\"\/collections\/nvidia-dgx-spark\"\u003eNVIDIA DGX Spark \u0026amp; DGX Station collection\u003c\/a\u003e.\u003c\/p\u003e\n\u003c\/div\u003e\u003c\/details\u003e\n\u003cdetails class=\"pd-acc\"\u003e\u003csummary style=\"cursor: pointer; font-weight: 700;\"\u003eDatasheet \u0026amp; Downloads\u003c\/summary\u003e\u003cdiv class=\"pd-acc-body\"\u003e\n\u003cul\u003e\n\u003cli\u003e\u003ca href=\"https:\/\/lenovopress.lenovo.com\/lp2357-lenovo-nvidia-gb300-nvl72-rack-scale-ai\" target=\"_blank\" rel=\"noopener\"\u003eLenovo NVIDIA GB300 NVL72 Product Guide (Lenovo Press LP2357)\u003c\/a\u003e\u003c\/li\u003e\n\u003cli\u003e\u003ca href=\"https:\/\/lenovopress.lenovo.com\/lp2357.pdf\" target=\"_blank\" rel=\"noopener\"\u003eProduct Guide (PDF)\u003c\/a\u003e\u003c\/li\u003e\n\u003cli\u003e\u003ca href=\"https:\/\/lenovopress.lenovo.com\/datasheet\/DS0207\" target=\"_blank\" rel=\"noopener\"\u003eLenovo NVIDIA GB300 NVL72 datasheet\u003c\/a\u003e\u003c\/li\u003e\n\u003cli\u003e\u003ca href=\"https:\/\/lenovopress.lenovo.com\/3dtours\/gb300nvl72\" target=\"_blank\" rel=\"noopener\"\u003eInteractive 3D tour\u003c\/a\u003e\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003c\/div\u003e\u003c\/details\u003e\n\u003cdetails class=\"pd-acc\"\u003e\u003csummary style=\"cursor: pointer; font-weight: 700;\"\u003eFrequently Asked Questions - Lenovo NVIDIA GB300 NVL72\u003c\/summary\u003e\u003cdiv class=\"pd-acc-body\"\u003e\n\u003ch3\u003eWhat is the Lenovo NVIDIA GB300 NVL72?\u003c\/h3\u003e\n\u003cp\u003eA liquid-cooled 48U rack with 72 NVIDIA Blackwell Ultra GPUs and 36 NVIDIA Grace CPUs in one NVLink domain, integrated and tested by Lenovo.\u003c\/p\u003e\n\u003ch3\u003eHow much power does one rack need?\u003c\/h3\u003e\n\u003cp\u003e135 kW TDP and up to 155 kW peak, fed by six or eight 33 kW power shelves.\u003c\/p\u003e\n\u003ch3\u003eDoes it need facility water?\u003c\/h3\u003e\n\u003cp\u003eYes. About 90% of the heat is removed by liquid, with inlet water up to 45 °C.\u003c\/p\u003e\n\u003ch3\u003eCan I buy a single compute tray?\u003c\/h3\u003e\n\u003cp\u003eNo. The compute and switch trays work only inside the GB300 NVL72 rack, which is sold as a complete system. Spare trays are quoted for existing racks.\u003c\/p\u003e\n\u003ch3\u003eHow do I buy it in Singapore?\u003c\/h3\u003e\n\u003cp\u003eEmail sales@sourceit.com.sg. SourceIT will arrange sizing, site readiness and a formal quotation.\u003c\/p\u003e\n\u003c\/div\u003e\u003c\/details\u003e\n\u003c\/div\u003e","brand":"Lenovo","offers":[{"title":"Default Title","offer_id":57615450570916,"sku":"7DJVCTO2WW","price":6000000.0,"currency_code":"SGD","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0605\/0136\/0804\/files\/lenovo-nvidia-gb300-nvl72-rack-1.jpg?v=1794474474"},{"product_id":"nvidia-gb300-nvl72-by-hpe-rack-scale-ai-supercomputer-72-blackwell-ultra-gpus-36-grace-cpus-direct-liquid-cooled","title":"NVIDIA GB300 NVL72 by HPE Rack-Scale AI Supercomputer - 72 Blackwell Ultra GPUs, 36 Grace CPUs, Direct Liquid Cooled","description":"\u003cdiv\u003e\n\u003cstyle\u003e\n.pd-acc { border: 1px solid #e2e2e2; border-radius: 8px; margin: 10px 0; overflow: hidden; }\n.pd-acc summary { cursor: pointer; padding: 14px 16px; font-weight: 700; font-size: 1.05em; list-style: none; display: flex; justify-content: space-between; align-items: center; background: #f7f7f7; }\n.pd-acc[open] \u003e summary { border-bottom: 1px solid #e2e2e2; }\n.pd-acc summary::-webkit-details-marker { display: none; }\n.pd-acc summary::after { content: \"+\"; font-size: 1.4em; font-weight: 400; line-height: 1; margin-left: 12px; }\n.pd-acc[open] \u003e summary::after { content: \"\\2212\"; }\n.pd-acc .pd-acc-body { padding: 8px 16px 14px; }\n.pd-spec, .pd-spec tr, .pd-spec td { border: none !important; background: transparent !important; }\n.pd-spec { width: 100%; border-collapse: collapse !important; margin: 6px 0 18px; }\n.pd-spec td { padding: 7px 10px 7px 0 !important; border-bottom: 1px solid #ececec !important; vertical-align: top; font-size: 0.95em; text-align: left; }\n.pd-spec td:first-child { width: 38%; font-weight: 600; color: #333; }\n.pd-spec tr:last-child td { border-bottom: none !important; }\n\u003c\/style\u003e\n\u003ch2\u003eNVIDIA GB300 NVL72 by HPE Rack-Scale AI Supercomputer, 72 NVIDIA Blackwell Ultra GPUs, 36 NVIDIA Grace CPUs, NVLink, HPE Direct Liquid Cooling, HPE Services\u003c\/h2\u003e\n\u003ch3\u003eRack-Scale AI for Models of Over 1 Trillion Parameters\u003c\/h3\u003e\n\u003cp\u003eNVIDIA GB300 NVL72 by HPE is a liquid-cooled, fully integrated rack-scale system. Its 72 NVIDIA Blackwell Ultra GPUs are interconnected over NVIDIA NVLink into a single 72-GPU domain that acts as one massive GPU, built for training and real-time inference on AI models of over 1 trillion parameters.\u003c\/p\u003e\n\u003cp\u003eSourceIT supplies NVIDIA GB300 NVL72 by HPE in Singapore as a project sale: site readiness, configuration, delivery and HPE services are scoped and quoted per deployment.\u003c\/p\u003e\n\u003ch3\u003eHPE Direct Liquid Cooling Expertise\u003c\/h3\u003e\n\u003cp\u003eHPE has long experience with large direct liquid cooling (DLC) deployments, and the GB300 NVL72 rack is delivered with HPE DLC technology so a dense AI rack can be operated with confidence.\u003c\/p\u003e\n\u003ch3\u003eFaster Time to Production with HPE Services\u003c\/h3\u003e\n\u003cp\u003eHPE services, backed by the HPE and NVIDIA partnership, cover the deployment of large AI clusters from installation to ongoing support.\u003c\/p\u003e\n\u003ch3\u003eThe NVIDIA GB300 NVL72 Platform\u003c\/h3\u003e\n\u003cp\u003eThe NVIDIA GB300 NVL72 reference design combines 72 Blackwell Ultra GPUs and 36 Arm-based NVIDIA Grace CPUs across 18 compute trays, with nine NVLink switch trays providing 1.8 TB\/s GPU-to-GPU links. NVIDIA rates it at 1.5x the AI performance of the GB200 NVL72.\u003c\/p\u003e\n\u003cdetails class=\"pd-acc\"\u003e\u003csummary style=\"cursor: pointer; font-weight: 700;\"\u003eNVIDIA GB300 NVL72 by HPE Datasheet - Technical Specifications\u003c\/summary\u003e\u003cdiv class=\"pd-acc-body\"\u003e\n\u003ch3\u003eGeneral\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eProduct Name\u003c\/td\u003e\n\u003ctd\u003eNVIDIA GB300 NVL72 by HPE\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eCategory\u003c\/td\u003e\n\u003ctd\u003eHPE AI Compute, rack-scale system\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003ePlatform\u003c\/td\u003e\n\u003ctd\u003eNVIDIA GB300 NVL72 (Grace Blackwell Ultra)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eCooling\u003c\/td\u003e\n\u003ctd\u003eHPE direct liquid cooling (DLC)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eServices\u003c\/td\u003e\n\u003ctd\u003eHPE Services for AI cluster deployment and support\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eWarranty\u003c\/td\u003e\n\u003ctd\u003ePer HPE quotation and support contract\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\u003ch3\u003eCompute (NVIDIA GB300 NVL72 platform)\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eGPUs per Rack\u003c\/td\u003e\n\u003ctd\u003e72 x NVIDIA Blackwell Ultra\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eCPUs per Rack\u003c\/td\u003e\n\u003ctd\u003e36 x NVIDIA Grace (Arm Neoverse V2, 72 cores each)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eCompute Trays\u003c\/td\u003e\n\u003ctd\u003e18 x 1U, each with 2 Grace CPUs and 4 Blackwell Ultra GPUs\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eNVLink Switch Trays\u003c\/td\u003e\n\u003ctd\u003e9\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eGPU Memory\u003c\/td\u003e\n\u003ctd\u003e288 GB HBM3e per GPU\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eGPU-to-GPU Interconnect\u003c\/td\u003e\n\u003ctd\u003eFifth-generation NVIDIA NVLink, 1.8 TB\/s\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eModel Scale\u003c\/td\u003e\n\u003ctd\u003eAI models of over 1 trillion parameters (HPE)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\u003ch3\u003eRack, Power and Networking\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eRack Dimensions and Weight\u003c\/td\u003e\n\u003ctd\u003eConfirmed with the HPE configuration at quotation\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003ePower\u003c\/td\u003e\n\u003ctd\u003eRack-level busbar power; power shelf count and redundancy confirmed at quotation\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eScale-Out Networking\u003c\/td\u003e\n\u003ctd\u003eNVIDIA networking options confirmed at quotation\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\u003c\/div\u003e\u003c\/details\u003e\n\u003cdetails class=\"pd-acc\"\u003e\u003csummary style=\"cursor: pointer; font-weight: 700;\"\u003eCompatibility \u0026amp; Software Support\u003c\/summary\u003e\u003cdiv class=\"pd-acc-body\"\u003e\n\u003ch3\u003eNVIDIA AI Software\u003c\/h3\u003e\n\u003cp\u003eThe system runs the NVIDIA accelerated computing stack, including CUDA, NVIDIA AI Enterprise, NIM microservices and NGC containers for arm64.\u003c\/p\u003e\n\u003ch3\u003eArm64 Architecture\u003c\/h3\u003e\n\u003cp\u003eThe Grace CPU is an Arm Neoverse V2 processor, so operating system and container images are arm64 builds. Check any x86-only commercial software before ordering.\u003c\/p\u003e\n\u003ch3\u003eRelated HPE AI Systems\u003c\/h3\u003e\n\u003cp\u003eHPE also offers the NVIDIA GB200 NVL72 by HPE, HPE ProLiant Compute XD685 and HPE Compute XD690 for other AI cluster designs. Ask SourceIT for a comparison.\u003c\/p\u003e\n\u003c\/div\u003e\u003c\/details\u003e\n\u003cdetails class=\"pd-acc\"\u003e\u003csummary style=\"cursor: pointer; font-weight: 700;\"\u003eSite Requirements \u0026amp; Ordering Information\u003c\/summary\u003e\u003cdiv class=\"pd-acc-body\"\u003e\n\u003ch3\u003eWhat to Plan For\u003c\/h3\u003e\n\u003cul\u003e\n\u003cli\u003eHigh-density three-phase power for a rack in the 130 kW class\u003c\/li\u003e\n\u003cli\u003eFacility water loop or CDU capacity for direct liquid cooling\u003c\/li\u003e\n\u003cli\u003eFloor loading, delivery path and lift access for a fully populated rack\u003c\/li\u003e\n\u003cli\u003eCluster fabric, storage and management networks\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003eHow This Is Sold\u003c\/h3\u003e\n\u003cp\u003eThe price shown is an indicative price for one rack. Final pricing depends on configuration and services and is confirmed by formal quotation. Email \u003ca href=\"mailto:sales@sourceit.com.sg\"\u003esales@sourceit.com.sg\u003c\/a\u003e for configuration, lead time and a GST quotation.\u003c\/p\u003e\n\u003cp\u003eFor deskside AI systems, see the \u003ca href=\"\/collections\/nvidia-dgx-spark\"\u003eNVIDIA DGX Spark \u0026amp; DGX Station collection\u003c\/a\u003e.\u003c\/p\u003e\n\u003c\/div\u003e\u003c\/details\u003e\n\u003cdetails class=\"pd-acc\"\u003e\u003csummary style=\"cursor: pointer; font-weight: 700;\"\u003eDatasheet \u0026amp; Downloads\u003c\/summary\u003e\u003cdiv class=\"pd-acc-body\"\u003e\n\u003cul\u003e\n\u003cli\u003e\u003ca href=\"https:\/\/www.hpe.com\/us\/en\/products\/compute\/hpe-nvidia-gb300-nvl72.html\" target=\"_blank\" rel=\"noopener\"\u003eNVIDIA GB300 NVL72 by HPE - product page\u003c\/a\u003e\u003c\/li\u003e\n\u003cli\u003e\u003ca href=\"https:\/\/www.hpe.com\/psnow\/doc\/a50014019enw\" target=\"_blank\" rel=\"noopener\"\u003eSolution overview\u003c\/a\u003e\u003c\/li\u003e\n\u003cli\u003e\u003ca href=\"https:\/\/www.hpe.com\/psnow\/doc\/a00146181enw\" target=\"_blank\" rel=\"noopener\"\u003eInfographic\u003c\/a\u003e\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003c\/div\u003e\u003c\/details\u003e\n\u003cdetails class=\"pd-acc\"\u003e\u003csummary style=\"cursor: pointer; font-weight: 700;\"\u003eFrequently Asked Questions - NVIDIA GB300 NVL72 by HPE\u003c\/summary\u003e\u003cdiv class=\"pd-acc-body\"\u003e\n\u003ch3\u003eWhat is NVIDIA GB300 NVL72 by HPE?\u003c\/h3\u003e\n\u003cp\u003eA liquid-cooled, fully integrated rack with 72 NVIDIA Blackwell Ultra GPUs in one NVLink domain, delivered with HPE direct liquid cooling and HPE Services.\u003c\/p\u003e\n\u003ch3\u003eWhat workloads is it for?\u003c\/h3\u003e\n\u003cp\u003eTraining and real-time inference for very large AI models, including models of over 1 trillion parameters.\u003c\/p\u003e\n\u003ch3\u003eDoes it need facility water?\u003c\/h3\u003e\n\u003cp\u003eYes. It is a direct liquid cooled system and needs a facility water loop or CDU capacity.\u003c\/p\u003e\n\u003ch3\u003eHow do I buy it in Singapore?\u003c\/h3\u003e\n\u003cp\u003eEmail sales@sourceit.com.sg. SourceIT will arrange sizing, site readiness and a formal quotation.\u003c\/p\u003e\n\u003c\/div\u003e\u003c\/details\u003e\n\u003ch3\u003eProduct Video\u003c\/h3\u003e\n\u003cp\u003e\u003ciframe width=\"560\" height=\"315\" src=\"https:\/\/www.youtube.com\/embed\/P9D3p3n9mUw\" title=\"Unleash unprecedented AI acceleration with NVIDIA GB300 NVL72 by HPE\" frameborder=\"0\" allow=\"accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture\" allowfullscreen\u003e\u003c\/iframe\u003e\u003c\/p\u003e\n\u003c\/div\u003e","brand":"HPE","offers":[{"title":"Default Title","offer_id":57615450603684,"sku":"HPE-GB300-NVL72","price":6000000.0,"currency_code":"SGD","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0605\/0136\/0804\/files\/hpe-nvidia-gb300-nvl72-rack-front-v2.png?v=1794554091"},{"product_id":"asus-ai-pod-xa-gb721-e2-nvidia-gb300-nvl72-rack-scale-ai-supercomputer-72-blackwell-ultra-gpus-36-grace-cpus-liquid-cooled","title":"ASUS AI POD XA GB721-E2 NVIDIA GB300 NVL72 Rack-Scale AI Supercomputer - 72 Blackwell Ultra GPUs, 36 Grace CPUs, Liquid Cooled","description":"\u003cdiv\u003e\n\u003cstyle\u003e\n.pd-acc { border: 1px solid #e2e2e2; border-radius: 8px; margin: 10px 0; overflow: hidden; }\n.pd-acc summary { cursor: pointer; padding: 14px 16px; font-weight: 700; font-size: 1.05em; list-style: none; display: flex; justify-content: space-between; align-items: center; background: #f7f7f7; }\n.pd-acc[open] \u003e summary { border-bottom: 1px solid #e2e2e2; }\n.pd-acc summary::-webkit-details-marker { display: none; }\n.pd-acc summary::after { content: \"+\"; font-size: 1.4em; font-weight: 400; line-height: 1; margin-left: 12px; }\n.pd-acc[open] \u003e summary::after { content: \"\\2212\"; }\n.pd-acc .pd-acc-body { padding: 8px 16px 14px; }\n.pd-spec, .pd-spec tr, .pd-spec td { border: none !important; background: transparent !important; }\n.pd-spec { width: 100%; border-collapse: collapse !important; margin: 6px 0 18px; }\n.pd-spec td { padding: 7px 10px 7px 0 !important; border-bottom: 1px solid #ececec !important; vertical-align: top; font-size: 0.95em; text-align: left; }\n.pd-spec td:first-child { width: 38%; font-weight: 600; color: #333; }\n.pd-spec tr:last-child td { border-bottom: none !important; }\n\u003c\/style\u003e\n\u003ch2\u003eASUS AI POD XA GB721-E2 NVIDIA GB300 NVL72 Rack-Scale AI Supercomputer, 72 NVIDIA Blackwell Ultra GPUs, 36 NVIDIA Grace CPUs, 5th-Gen NVLink, ConnectX-8 SuperNIC, BlueField-3 DPU, 48U MGX Rack, Liquid Cooled (XA GB721-E2)\u003c\/h2\u003e\n\u003ch3\u003eASUS AI POD with NVIDIA GB300 NVL72: Built for AI Reasoning\u003c\/h3\u003e\n\u003cp\u003eThe ASUS XA GB721-E2 is the ASUS AI POD full-rack system built on NVIDIA GB300 NVL72. It combines 72 NVIDIA Blackwell Ultra GPUs and 36 NVIDIA Grace CPUs in a liquid-cooled, 48U NVIDIA MGX-compliant rack with 18 compute trays and nine NVLink switch trays, engineered for trillion-parameter AI reasoning, LLM training and inference.\u003c\/p\u003e\n\u003cp\u003eSourceIT supplies the ASUS XA GB721-E2 in Singapore as a project sale: site readiness, configuration, delivery and ASUS professional services are scoped and quoted per deployment.\u003c\/p\u003e\n\u003ch3\u003eOne NVLink Domain at 1.8 TB\/s\u003c\/h3\u003e\n\u003cp\u003eFifth-generation NVIDIA NVLink links every GPU in the rack at 1.8 TB\/s, so 72 GPUs work as one unified, high-bandwidth fabric. NVIDIA rates GB300 NVL72 at 1.5x the AI performance of GB200 NVL72.\u003c\/p\u003e\n\u003ch3\u003eConnectX-8 SuperNICs and BlueField-3 DPUs\u003c\/h3\u003e\n\u003cp\u003eNVIDIA ConnectX-8 SuperNICs with NVIDIA Quantum-X800 InfiniBand or Spectrum-X Ethernet provide low-latency scale-out to multi-rack clusters, with BlueField-3 DPUs for accelerated data movement.\u003c\/p\u003e\n\u003ch3\u003eLiquid-to-Air or Liquid-to-Liquid Cooling\u003c\/h3\u003e\n\u003cp\u003eManifold-based liquid cooling is offered in liquid-to-air and liquid-to-liquid designs, so the rack can be matched to data halls with or without facility water.\u003c\/p\u003e\n\u003ch3\u003eRapid Deployment with ACC and AIDC\u003c\/h3\u003e\n\u003cp\u003eASUS Control Center (ACC) and ASUS Infrastructure Deployment Center (AIDC) provide zero-touch onboarding, centralized configuration and rack-scale deployment automation, and ASUS backs the rack with end-to-end design, integration and lifecycle services.\u003c\/p\u003e\n\u003cdetails class=\"pd-acc\"\u003e\u003csummary style=\"cursor: pointer; font-weight: 700;\"\u003eASUS XA GB721-E2 Datasheet - Technical Specifications\u003c\/summary\u003e\u003cdiv class=\"pd-acc-body\"\u003e\n\u003ch3\u003eGeneral\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eProduct Name\u003c\/td\u003e\n\u003ctd\u003eASUS AI POD with NVIDIA GB300 NVL72\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eModel\u003c\/td\u003e\n\u003ctd\u003eXA GB721-E2\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003ePlatform\u003c\/td\u003e\n\u003ctd\u003eNVIDIA GB300 NVL72 (Grace Blackwell Ultra)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eForm Factor\u003c\/td\u003e\n\u003ctd\u003e48U NVIDIA MGX-compliant rack\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eWarranty\u003c\/td\u003e\n\u003ctd\u003ePer ASUS quotation and support contract\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\u003ch3\u003eCompute\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eGPUs per Rack\u003c\/td\u003e\n\u003ctd\u003e72 x NVIDIA Blackwell Ultra\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eCPUs per Rack\u003c\/td\u003e\n\u003ctd\u003e36 x NVIDIA Grace\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eCompute Trays\u003c\/td\u003e\n\u003ctd\u003e18\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eNVLink Switch Trays\u003c\/td\u003e\n\u003ctd\u003e9\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eGPU-to-GPU Interconnect\u003c\/td\u003e\n\u003ctd\u003eFifth-generation NVIDIA NVLink, 1.8 TB\/s\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eServiceability\u003c\/td\u003e\n\u003ctd\u003eBlackwell Ultra GPU SXM7 and SOCAMM module design\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\u003ch3\u003eNetworking\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eSuperNIC\u003c\/td\u003e\n\u003ctd\u003eNVIDIA ConnectX-8, 800 Gb\/s per GPU\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eDPU\u003c\/td\u003e\n\u003ctd\u003eNVIDIA BlueField-3\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eCluster Fabric\u003c\/td\u003e\n\u003ctd\u003eNVIDIA Quantum-X800 InfiniBand or Spectrum-X Ethernet\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\u003ch3\u003eCooling, Power and Management\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eCooling\u003c\/td\u003e\n\u003ctd\u003eManifold-based liquid cooling; liquid-to-air or liquid-to-liquid\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003ePower\u003c\/td\u003e\n\u003ctd\u003eRack-level busbar power; power shelf count and redundancy confirmed at quotation\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eManagement Software\u003c\/td\u003e\n\u003ctd\u003eASUS Control Center (ACC), ASUS Infrastructure Deployment Center (AIDC)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eRack Dimensions and Weight\u003c\/td\u003e\n\u003ctd\u003eSee ASUS datasheet; confirmed at quotation\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\u003c\/div\u003e\u003c\/details\u003e\n\u003cdetails class=\"pd-acc\"\u003e\u003csummary style=\"cursor: pointer; font-weight: 700;\"\u003eCompatibility \u0026amp; Software Support\u003c\/summary\u003e\u003cdiv class=\"pd-acc-body\"\u003e\n\u003ch3\u003eNVIDIA AI Software\u003c\/h3\u003e\n\u003cp\u003eThe system runs the NVIDIA accelerated computing stack, including CUDA, NVIDIA AI Enterprise, NIM microservices and NGC containers for arm64.\u003c\/p\u003e\n\u003ch3\u003eArm64 Architecture\u003c\/h3\u003e\n\u003cp\u003eThe Grace CPU is an Arm Neoverse V2 processor, so operating system and container images are arm64 builds. Check any x86-only commercial software before ordering.\u003c\/p\u003e\n\u003ch3\u003eASUS AI POD Ecosystem\u003c\/h3\u003e\n\u003cp\u003eASUS offers storage, networking, cooling and management as part of the AI POD, using validated reference architectures for predictable performance and simpler scaling.\u003c\/p\u003e\n\u003c\/div\u003e\u003c\/details\u003e\n\u003cdetails class=\"pd-acc\"\u003e\u003csummary style=\"cursor: pointer; font-weight: 700;\"\u003eSite Requirements \u0026amp; Ordering Information\u003c\/summary\u003e\u003cdiv class=\"pd-acc-body\"\u003e\n\u003ch3\u003eWhat to Plan For\u003c\/h3\u003e\n\u003cul\u003e\n\u003cli\u003eHigh-density three-phase power for a rack in the 130 kW class\u003c\/li\u003e\n\u003cli\u003eFacility water loop for liquid-to-liquid cooling, or space and airflow for liquid-to-air units\u003c\/li\u003e\n\u003cli\u003eFloor loading, delivery path and lift access for a fully populated 48U rack\u003c\/li\u003e\n\u003cli\u003eCluster fabric, storage and management networks\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003eHow This Is Sold\u003c\/h3\u003e\n\u003cp\u003eThe price shown is an indicative price for one rack. Final pricing depends on configuration and services and is confirmed by formal quotation. Email \u003ca href=\"mailto:sales@sourceit.com.sg\"\u003esales@sourceit.com.sg\u003c\/a\u003e for configuration, lead time and a GST quotation.\u003c\/p\u003e\n\u003cp\u003eFor deskside AI systems, see the \u003ca href=\"\/collections\/nvidia-dgx-spark\"\u003eNVIDIA DGX Spark \u0026amp; DGX Station collection\u003c\/a\u003e.\u003c\/p\u003e\n\u003c\/div\u003e\u003c\/details\u003e\n\u003cdetails class=\"pd-acc\"\u003e\u003csummary style=\"cursor: pointer; font-weight: 700;\"\u003eDatasheet \u0026amp; Downloads\u003c\/summary\u003e\u003cdiv class=\"pd-acc-body\"\u003e\n\u003cul\u003e\n\u003cli\u003e\u003ca href=\"https:\/\/servers.asus.com\/products\/asus-ai-pod\/nvidia-grace-blackwell\/XA-GB721-E2\" target=\"_blank\" rel=\"noopener\"\u003eASUS XA GB721-E2 - product page\u003c\/a\u003e\u003c\/li\u003e\n\u003cli\u003e\u003ca href=\"https:\/\/dlcdnets.asus.com\/pub\/ASUS\/server\/XA%20GB721-E2\/Datasheet\/Datasheet_XA_GB721-E2_NVIDIA_GB300_NVL72_V1.pdf\" target=\"_blank\" rel=\"noopener\"\u003eASUS XA GB721-E2 datasheet (PDF)\u003c\/a\u003e\u003c\/li\u003e\n\u003cli\u003e\u003ca href=\"https:\/\/www.asus.com\/event\/nvidia-gb300-nvl72\" target=\"_blank\" rel=\"noopener\"\u003eASUS AI POD with NVIDIA GB300 NVL72 - overview\u003c\/a\u003e\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003c\/div\u003e\u003c\/details\u003e\n\u003cdetails class=\"pd-acc\"\u003e\u003csummary style=\"cursor: pointer; font-weight: 700;\"\u003eFrequently Asked Questions - ASUS XA GB721-E2\u003c\/summary\u003e\u003cdiv class=\"pd-acc-body\"\u003e\n\u003ch3\u003eWhat is the ASUS XA GB721-E2?\u003c\/h3\u003e\n\u003cp\u003eThe ASUS AI POD rack built on NVIDIA GB300 NVL72: 72 Blackwell Ultra GPUs and 36 Grace CPUs in a liquid-cooled 48U rack.\u003c\/p\u003e\n\u003ch3\u003eHow is it different from the ASUS ESC NM2N721-E1?\u003c\/h3\u003e\n\u003cp\u003eThe ESC NM2N721-E1 is the earlier GB200 NVL72 rack with Blackwell GPUs. The XA GB721-E2 uses Blackwell Ultra GPUs on the GB300 NVL72 platform.\u003c\/p\u003e\n\u003ch3\u003eDoes it need facility water?\u003c\/h3\u003e\n\u003cp\u003eNot always. ASUS offers liquid-to-liquid cooling for sites with facility water and liquid-to-air cooling for sites without.\u003c\/p\u003e\n\u003ch3\u003eHow do I buy it in Singapore?\u003c\/h3\u003e\n\u003cp\u003eEmail sales@sourceit.com.sg. SourceIT will arrange sizing, site readiness and a formal quotation.\u003c\/p\u003e\n\u003c\/div\u003e\u003c\/details\u003e\n\u003ch3\u003eProduct Video\u003c\/h3\u003e\n\u003cp\u003e\u003ciframe width=\"560\" height=\"315\" src=\"https:\/\/www.youtube.com\/embed\/lZVpYKY95CE\" title=\"NVIDIA GB300 NVL72 | ASUS AI POD\" frameborder=\"0\" allow=\"accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture\" allowfullscreen\u003e\u003c\/iframe\u003e\u003c\/p\u003e\n\u003c\/div\u003e","brand":"Asus","offers":[{"title":"Default Title","offer_id":57615450734756,"sku":"XA GB721-E2","price":6000000.0,"currency_code":"SGD","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0605\/0136\/0804\/files\/w1000_01bc467a-4197-45a5-b510-c711b9ec74bc.png?v=1794473457"},{"product_id":"dell-poweredge-xe9712-nvidia-gb300-nvl72-rack-scale-ai-supercomputer-72-blackwell-ultra-gpus-36-grace-cpus-liquid-cooled","title":"Dell PowerEdge XE9712 NVIDIA GB300 NVL72 Rack-Scale AI Supercomputer - 72 Blackwell Ultra GPUs, 36 Grace CPUs, Liquid Cooled","description":"\u003cdiv\u003e\n\u003cstyle\u003e\n.pd-acc { border: 1px solid #e2e2e2; border-radius: 8px; margin: 10px 0; overflow: hidden; }\n.pd-acc summary { cursor: pointer; padding: 14px 16px; font-weight: 700; font-size: 1.05em; list-style: none; display: flex; justify-content: space-between; align-items: center; background: #f7f7f7; }\n.pd-acc[open] \u003e summary { border-bottom: 1px solid #e2e2e2; }\n.pd-acc summary::-webkit-details-marker { display: none; }\n.pd-acc summary::after { content: \"+\"; font-size: 1.4em; font-weight: 400; line-height: 1; margin-left: 12px; }\n.pd-acc[open] \u003e summary::after { content: \"\\2212\"; }\n.pd-acc .pd-acc-body { padding: 8px 16px 14px; }\n.pd-spec, .pd-spec tr, .pd-spec td { border: none !important; background: transparent !important; }\n.pd-spec { width: 100%; border-collapse: collapse !important; margin: 6px 0 18px; }\n.pd-spec td { padding: 7px 10px 7px 0 !important; border-bottom: 1px solid #ececec !important; vertical-align: top; font-size: 0.95em; text-align: left; }\n.pd-spec td:first-child { width: 38%; font-weight: 600; color: #333; }\n.pd-spec tr:last-child td { border-bottom: none !important; }\n\u003c\/style\u003e\n\u003ch2\u003eDell PowerEdge XE9712 NVIDIA GB300 NVL72 Rack-Scale AI Supercomputer, 72 NVIDIA Blackwell Ultra GPUs, 36 NVIDIA Grace CPUs, NVLink 1.8 TB\/s, Dell IR9048 48U Integrated Rack, Direct Liquid Cooling (XE9712)\u003c\/h2\u003e\n\u003ch3\u003eDell Integrated Rack with NVIDIA GB300 NVL72\u003c\/h3\u003e\n\u003cp\u003eThe Dell PowerEdge XE9712 is a rack-scale server platform built on the NVIDIA Grace Blackwell Ultra GB300 NVL72 architecture. Delivered in the Dell IR9048 integrated rack, it unifies up to 72 NVIDIA Blackwell Ultra GPUs and 36 Arm-based NVIDIA Grace CPUs over NVLink with 1.8 TB\/s of GPU-to-GPU bandwidth, for foundation model training and real-time inference at scale.\u003c\/p\u003e\n\u003cp\u003eSourceIT supplies the PowerEdge XE9712 rack in Singapore as a project sale: site readiness, configuration, delivery and Dell services are scoped and quoted per deployment.\u003c\/p\u003e\n\u003ch3\u003eUp to 17 TB of CPU Memory and 288 GB HBM3e per GPU\u003c\/h3\u003e\n\u003cp\u003eEach 1U XE9712 sled holds two 72-core Grace CPUs with 480 GB of LPDDR5X ECC memory each, and four Blackwell Ultra GPUs with 288 GB of HBM3e each, linked by a 900 GB\/s coherent CPU-GPU NVLink connection.\u003c\/p\u003e\n\u003ch3\u003eDell Integrated Rack Scalable Systems\u003c\/h3\u003e\n\u003cp\u003eDell's Integrated Rack Scalable Systems (IRSS) program delivers the IR9048 as a pre-built, tested rack. The Open Rack v3 design feeds sleds from 33 kW power shelves over a busbar, removing in-rack PDUs, and uses direct liquid cooling for the high-power GPUs and CPUs.\u003c\/p\u003e\n\u003ch3\u003eConnectX-8 Networking and E1.S NVMe Storage\u003c\/h3\u003e\n\u003cp\u003eEach sled has four OSFP ports on ConnectX-8, a BlueField-3 option, and up to eight hot-swap E1.S NVMe drives for local data.\u003c\/p\u003e\n\u003ch3\u003eOpenManage and iDRAC Management\u003c\/h3\u003e\n\u003cp\u003eSleds are available with OpenBMC or iDRAC 10, each linked to the NVIDIA HMC, and integrate with Dell OpenManage for fleet operations, security and automation.\u003c\/p\u003e\n\u003cdetails class=\"pd-acc\"\u003e\u003csummary style=\"cursor: pointer; font-weight: 700;\"\u003eDell PowerEdge XE9712 \/ IR9048 Datasheet - Technical Specifications\u003c\/summary\u003e\u003cdiv class=\"pd-acc-body\"\u003e\n\u003ch3\u003eGeneral\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eProduct Name\u003c\/td\u003e\n\u003ctd\u003eDell PowerEdge XE9712 with NVIDIA GB300 NVL72\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eRack\u003c\/td\u003e\n\u003ctd\u003eDell IR9048 integrated rack, ORv3 (OCP Open Rack v3), 48 RU\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003ePlatform\u003c\/td\u003e\n\u003ctd\u003eNVIDIA GB300 NVL72 (Grace Blackwell Ultra)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eGPUs per Rack\u003c\/td\u003e\n\u003ctd\u003eUp to 72 x NVIDIA Blackwell Ultra\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eCPUs per Rack\u003c\/td\u003e\n\u003ctd\u003e36 x NVIDIA Grace\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eSupport\u003c\/td\u003e\n\u003ctd\u003eDell ProSupport or ProSupport Plus recommended; per Dell quotation\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\u003ch3\u003eXE9712 Sled (per 1 RU sled)\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eProcessors\u003c\/td\u003e\n\u003ctd\u003e2 x NVIDIA Grace, 72 cores each\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eGPUs\u003c\/td\u003e\n\u003ctd\u003e4 x NVIDIA Blackwell Ultra, 288 GB HBM3e each\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eCPU Memory\u003c\/td\u003e\n\u003ctd\u003e480 GB LPDDR5X ECC per CPU\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eCPU-GPU Link\u003c\/td\u003e\n\u003ctd\u003eNVLink, 900 GB\/s coherent\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eFront Drive Bays\u003c\/td\u003e\n\u003ctd\u003eUp to 8 x hot-swap EDSFF E1.S NVMe\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eBoot Drive\u003c\/td\u003e\n\u003ctd\u003e1 x M.2\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eNetworking\u003c\/td\u003e\n\u003ctd\u003e4 x OSFP (ConnectX-8), 1G integrated LOM\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003ePCIe\u003c\/td\u003e\n\u003ctd\u003eUp to 2 x FHHL Gen5 x16 (front risers), 2 x mezzanine Gen5\/Gen6, 1 x BlueField-3 SuperNIC\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eManagement\u003c\/td\u003e\n\u003ctd\u003eASPEED AST2600 BMC with OpenBMC, or iDRAC 10; BMC-to-NVIDIA HMC\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eSecurity\u003c\/td\u003e\n\u003ctd\u003eSilicon root of trust, protected supply chain, MFA, role-based access control\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eOperating System\u003c\/td\u003e\n\u003ctd\u003eCanonical Ubuntu Server LTS\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eSled Dimensions \/ Weight\u003c\/td\u003e\n\u003ctd\u003e43.6 x 438 x 836.8 mm (H x W x D); 30 kg\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\u003ch3\u003eIR9048 Rack, Power and Cooling\u003c\/h3\u003e\n\u003ctable class=\"pd-spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eGPU-to-GPU Interconnect\u003c\/td\u003e\n\u003ctd\u003eNVIDIA NVLink, 1.8 TB\/s\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003ePower Shelf\u003c\/td\u003e\n\u003ctd\u003e33 kW (PS33), 6 x 5500 W AC PSUs, up to 54 V DC over rack busbar\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eCooling\u003c\/td\u003e\n\u003ctd\u003eDirect liquid cooling plus air; 8 dual-rotor fans per sled\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eRack Dimensions\u003c\/td\u003e\n\u003ctd\u003e2294 x 750 x 1200 mm (H x W x D); 1340 mm deep with stabilizers\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eRack Weight\u003c\/td\u003e\n\u003ctd\u003e1590 kg (wet cabinet)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\u003c\/div\u003e\u003c\/details\u003e\n\u003cdetails class=\"pd-acc\"\u003e\u003csummary style=\"cursor: pointer; font-weight: 700;\"\u003eCompatibility \u0026amp; Software Support\u003c\/summary\u003e\u003cdiv class=\"pd-acc-body\"\u003e\n\u003ch3\u003eNVIDIA AI Software\u003c\/h3\u003e\n\u003cp\u003eThe system runs the NVIDIA accelerated computing stack, including CUDA, NVIDIA AI Enterprise, NIM microservices and NGC containers for arm64, and is part of the Dell AI Factory with NVIDIA.\u003c\/p\u003e\n\u003ch3\u003eArm64 Architecture\u003c\/h3\u003e\n\u003cp\u003eThe Grace CPU is an Arm Neoverse V2 processor, so operating system and container images are arm64 builds. Check any x86-only commercial software before ordering.\u003c\/p\u003e\n\u003ch3\u003eDell Networking\u003c\/h3\u003e\n\u003cp\u003eThe rack supports Dell PowerSwitch and NVIDIA networking for the cluster fabric. Switches, optics and cabling are scoped with the project.\u003c\/p\u003e\n\u003c\/div\u003e\u003c\/details\u003e\n\u003cdetails class=\"pd-acc\"\u003e\u003csummary style=\"cursor: pointer; font-weight: 700;\"\u003eSite Requirements \u0026amp; Ordering Information\u003c\/summary\u003e\u003cdiv class=\"pd-acc-body\"\u003e\n\u003ch3\u003eWhat to Plan For\u003c\/h3\u003e\n\u003cul\u003e\n\u003cli\u003eHigh-density three-phase power for multiple 33 kW power shelves\u003c\/li\u003e\n\u003cli\u003eFacility water loop or CDU capacity for direct liquid cooling\u003c\/li\u003e\n\u003cli\u003eFloor loading for a 1590 kg rack, plus delivery path and lift access\u003c\/li\u003e\n\u003cli\u003eCluster fabric, storage and management networks\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003eHow This Is Sold\u003c\/h3\u003e\n\u003cp\u003eThe price shown is an indicative price for one rack. The XE9712 sled works only inside the IR9048 rack and is sold as a rack-scale solution. Final pricing depends on configuration and services and is confirmed by formal quotation. Email \u003ca href=\"mailto:sales@sourceit.com.sg\"\u003esales@sourceit.com.sg\u003c\/a\u003e for configuration, lead time and a GST quotation.\u003c\/p\u003e\n\u003cp\u003eFor deskside AI systems, see the \u003ca href=\"\/collections\/nvidia-dgx-spark\"\u003eNVIDIA DGX Spark \u0026amp; DGX Station collection\u003c\/a\u003e.\u003c\/p\u003e\n\u003c\/div\u003e\u003c\/details\u003e\n\u003cdetails class=\"pd-acc\"\u003e\u003csummary style=\"cursor: pointer; font-weight: 700;\"\u003eDatasheet \u0026amp; Downloads\u003c\/summary\u003e\u003cdiv class=\"pd-acc-body\"\u003e\n\u003cul\u003e\n\u003cli\u003e\u003ca href=\"https:\/\/www.delltechnologies.com\/asset\/en-au\/products\/servers\/technical-support\/poweredge-xe9712-spec-sheet.pdf\" target=\"_blank\" rel=\"noopener\"\u003eDell PowerEdge XE9712 spec sheet (PDF)\u003c\/a\u003e\u003c\/li\u003e\n\u003cli\u003e\u003ca href=\"https:\/\/www.dell.com\/en-au\/shop\/ipovw\/poweredge-xe9712\" target=\"_blank\" rel=\"noopener\"\u003eDell PowerEdge XE9712 - product page\u003c\/a\u003e\u003c\/li\u003e\n\u003cli\u003e\u003ca href=\"https:\/\/www.dell.com\/en-us\/shop\/storage-servers-and-networking-for-business\/sf\/integrated-rack-scalable-systems\" target=\"_blank\" rel=\"noopener\"\u003eDell Integrated Rack Scalable Systems\u003c\/a\u003e\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003c\/div\u003e\u003c\/details\u003e\n\u003cdetails class=\"pd-acc\"\u003e\u003csummary style=\"cursor: pointer; font-weight: 700;\"\u003eFrequently Asked Questions - Dell PowerEdge XE9712\u003c\/summary\u003e\u003cdiv class=\"pd-acc-body\"\u003e\n\u003ch3\u003eWhat is the Dell PowerEdge XE9712?\u003c\/h3\u003e\n\u003cp\u003eDell's rack-scale AI platform on NVIDIA GB300 NVL72: 1U liquid-cooled sleds in the IR9048 rack, totalling up to 72 Blackwell Ultra GPUs and 36 Grace CPUs.\u003c\/p\u003e\n\u003ch3\u003eCan I buy a single XE9712 sled?\u003c\/h3\u003e\n\u003cp\u003eThe sled has no power supplies of its own and runs only in the IR9048 rack, so it is sold as a complete rack. Spare sleds are quoted for existing racks.\u003c\/p\u003e\n\u003ch3\u003eHow large and heavy is the rack?\u003c\/h3\u003e\n\u003cp\u003e2294 mm high, 750 mm wide and 1200 mm deep, with a wet cabinet weight of 1590 kg.\u003c\/p\u003e\n\u003ch3\u003eHow do I buy it in Singapore?\u003c\/h3\u003e\n\u003cp\u003eEmail sales@sourceit.com.sg. SourceIT will arrange sizing, site readiness and a formal quotation.\u003c\/p\u003e\n\u003c\/div\u003e\u003c\/details\u003e\n\u003c\/div\u003e","brand":"Dell","offers":[{"title":"Default Title","offer_id":57615450800292,"sku":"XE9712-GB300-NVL72","price":6000000.0,"currency_code":"SGD","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0605\/0136\/0804\/files\/dell-gb300-nvl72-rack-illustration-v2.png?v=1794556108"}],"url":"https:\/\/sourceit.com.sg\/collections\/nvidia-gb300-ai-supercomputers.oembed","provider":"SourceIT ","version":"1.0","type":"link"}